COMPARISION OF AERIAL IMAGERY AND SATELLITE IMAGERY FOR AUTONOMOUS VEHICLE PATH PLANNING

Size: px
Start display at page:

Download "COMPARISION OF AERIAL IMAGERY AND SATELLITE IMAGERY FOR AUTONOMOUS VEHICLE PATH PLANNING"

Transcription

1 8th International DAAAM Baltic Conference "INDUSTRIAL ENGINEERING April 2012, Tallinn, Estonia COMPARISION OF AERIAL IMAGERY AND SATELLITE IMAGERY FOR AUTONOMOUS VEHICLE PATH PLANNING Robert Hudjakov & Mart Tamre Abstract: The focus of our research is to improve navigation capabilities of an offroad unmanned ground vehicle (UGV) by improving its perception range. We train an ortophoto classifier by using data gathered by UGV and then extrapolate the data into wider area by using aerial imagery from unmanned aerial vehicles (UAV). The current progress report analyses feasibility of using satellite images instead of or together with aerial imagery and discusses an optimization technique that executes classifier as needed by path planner saving processing resources and time. Keywords: UGV, UAV, cost map, path planning. 1. INTRODUCTION There are battery powered off-road multipurpose UGV platforms developed in Tallinn University of Technology [1] objectives of which contains driving to target GPS coordinate, executing a task and returning to base. By fusing the on-board sensors of an UGV with overhead imagery we can increase the effective perception range of the UGV helping to reduce the time and energy requirements for executing said mission. In addition we can plan a path to objective ahead of time and predict the feasibility of the mission with current battery charge. Our proposed system consists of classifier, cost map generator and path planner. The system uses previously gathered UGV data and aerial imagery as an input and produces a long haul path for UGV, the prior knowledge and aerial imagery is used to train an image classifier that is then used for cost map generation and path planning. We intent to repeat the cycle of data gathering, classifier training, cost map generation and path planning after the mission is dispatched in order to keep the system flexible and to be able to consider new kinds of obstacles after they have been discovered by UGV. In this article we are analyzing the possibility of using satellite imagery together with or instead of aerial imagery. There are two explicit scenarios that we cover: whether it is possible to replace aerial imagery with lower quality satellite imagery and is it possible to switch over from aerial imagery to satellite imagery when UAV goes offline? The aerial imagery is preferred over satellite imagery for it has higher resolution and it can be gathered ahead or during the mission ensuring its freshness. In addition we describe an optimization technique that increases the performance of aerial imagery evaluation by combining cost map generation and path planning steps. The advantages are twofold. First we benefit from reduced processing requirements allowing cheaper and more mobile hardware (in terms of power requirements). Secondly it allows us to generate path faster which in turn allows us regenerate the updated path to

2 objective more often. 2. AERIAL IMAGERY VS. SATELLITE IMAGERY Thus far we have done all our experiments with aerial imagery and we have estimated that we get comparable results with satellite imagery, this time we decided to put the hypothesis to a bench. For the experiment we chose identical area to our aerial data set from Google maps database [2] [3] [4] [5], up-scaled the satellite imagery to identical resolution of aerial imagery and hand classified it (Illustration 1). It must be noted, however, that the satellite imagery has about 4x lower resolution, was taken during different time of the day, different time of the year and different year. The aerial data set was acquired during a summer time when grass was green and multiple construction projects were ongoing so there was a lot of construction debris. The satellite imagery was acquired a year later during spring when grass was not up yet and there was little debris from construction projects left. Our processing pipeline consists of three distinct steps: image classification, cost map generation and path planning. For image classification we pick patterns from aerial imagery and feed them into a classifier that outputs a feature vector. For the classification we use two approaches that are described in finer detail in our previous publication [5]: with method A we try to detect all features that are present on input pattern and with method B we try to detect to which category belongs the pixel in the middle of input pattern. The outputs of method A are independent meaning that the input pattern may contain multiple features that we are looking for (houses, roads, etc) while the outputs of method B are dependent - the center pixel may belong to only one category. The objectives of this experiment are twofold: We want to verify that our classifier can get comparable classification rates with satellite imagery. We want to verify the feasibility of cross usage of a trained classifier e.g. we want to see if a classifier that has been trained on aerial imagery is usable on satellite imagery and vice versa. Method A Method A is based on the idea that a convolutional artificial neural network is capable of extracting distinct features from input pattern and linear classifier layer can then be trained to report presence of objects in input pattern. In layman's terms each classifier output is trained to answer simple questions such as Is there a house in input pattern?, Is there a tree in input pattern or Is there some grass in input pattern?. All the classifier outputs are independent, there can be multiple features simultaneously present in the input pattern. For the experiment we used a classifier that is based on convolutional neural network and had three 29x29 neuron inputs for a 29x29pixel pattern with three color channels. The second hidden layer had six 13x13 neuron feature maps that were connected to all inputs using 5x5 neuron kernel and the third layer had fifty 5x5 neuron feature maps that were connected to all second layer feature maps using 5x5 neuron kernels. The final hidden layer consisted of a linear classifier with 100 neurons in it and the output layer had 4 neurons for four feature categories: houses, roads, grass and debris. The specific classifier configuration was chosen because we have reported multiple experiments with it in the past [2] [3] [4]. For the experiment we generated two data sets - one from aerial imagery and another from satellite imagery. Next we randomized the internal parameters of our classifier and trained it on aerial data set. Finally we tested the trained classifier on both aerial test set

3 and satellite test set. The same operation was repeated with a classifier trained on satellite data set and the results were saved into Table 1. Table 1 Aerial imagery classifier data set. Not only the color of the grass was different on the data sets, but also the sharp texture of grass on aerial imagery was replaced with botched texture of mud on Satellite imagery classifier Aerial data set Correctly classified: 97,9% Houses, rate=99,4% Roads, rate=99,7% Grass, rate=99,8% Debris, rate=98,8% Correctly classified: 11,2% Houses, rate=83% Roads, rate=78% Grass, rate=42% Debris, rate=50% Satellite data set Correctly classified: 9,4% Houses, rate=48% Roads, rate=87% Grass, rate=41% Debris, rate=48% As seen from above table the classifier is usable both with aerial imagery and satellite imagery - the classification rate in both cases is well over 90%. The cross use, however, is limited. The classifier that was trained on lower quality satellite imagery was capable of detecting houses on higher quality aerial imagery but the reverse did not hold. Correctly classified: 98,9% Houses, rate=99,7% Roads, rate=99,8% Grass, rate=99,9% Debris, rate=99,6% satellite imagery (Illustration 1). Similarly the aerial imagery had a lot of debris but satellite imagery had next to none and hence the cross use of classifiers for debris detection is not reliable. The roads were similar on both data sets and the classifier could detect the roads on both - this is important results because having a Illustration 1: Path planning using classifier with high uncertainty levels. Cost map (left) and planned path (right). Because on one imagery the grass was up and green while on other imagery it had not yet grown, the classifier trained on one data set can not be used to detect grass on other classifier that can detect roads is better than not having a classifier making our system usable even in unfavorable conditions.

4 Illustration 2: Grass on aerial imagery vs. mud on satellite imagery Our cost map generation algorithm is built to cope with classifier uncertainties and is capable of producing output even when classifier is only capable of detecting some of the features. By feeding the classifier trained on aerial imagery with satellite imagery we receive a cost map where everything besides roads are marked as unknown terrain and during navigation the roads are preferred. The following illustration (illustration 2) shows the path planning capability of our system with classifier that has been trained with aerial imagery using satellite imagery. Despite on the overall the classifier could only correctly classify 9,4% of patterns, it could find roads with relatively high certainty and thus can be used for path planning. The results are better with the classifier that is trained on satellite imagery and used on aerial imagery (Illustration 3). Both the roads and the buildings are clearly detected and since our proposition is to re-train the classifier while UGV is moving (using freshly gathered data) the classifier trained on satellite imagery has good starting values. To validate that claim we took a classifier that has been trained on satellite imagery and trained it for one single epoch on aerial imagery. The classification rate for all items went up from 11,2% to 66% and in detail the classification rate for houses went from 83% to 98%, for roads from 78% to 96%, for grass from 42% to 88% and for debris it went from 50% to 79%. We saw a similar increase after we took a classifier that had been trained on aerial imagery and used it on satellite imagery. After one epoch of training the total classification rate went up from 9,4% to 54%. Method B Method B will train the classifier to detect Illustration 3: Comparison of cost maps: satellite classifier on satellite imagery (left) vs. satellite classifier on aerial imagery (right). Darker areas have lower traversal cost than lighter areas.

5 the category of a single pixel in input image, particularly the pixel from the center of the pattern. The classifier will then answer the question Into which category belongs the pixel in the center of this pattern?. The categories are exclusive and thus the classifier outputs are dependent. The experiment setup was identical to method A: we generated two data sets and trained a classifier on each of them. The classifier were then tested and cross tested on the data sets and results are in the Table2. Table 2 Aerial imagery classifier Aerial data set Correctly classified: 88% Houses, rate=97,87% Roads, rate=97,07% Grass, rate=94,13% Debris, rate=97,93% Classifier trained on the aerial imagery works well on the the aerial imagery Classifier trained on the satellite imagery works well on the satellite imagery The classifier trained on lower quality imagery also works well on higher quality imagery but the reverse does not hold. The classifier that has been trained on satellite imagery is a good starting Satellite imagery classifier Correctly classified: 20,4% Houses, rate=66,4% Roads, rate=67,57% Grass, rate=76,23% Debris, rate=52% Satellite data set Correctly classified: 12,7% Houses, rate=55,73% Roads, rate=77,77% Grass, rate=48,67% Debris, rate=57,93% The results are very similar to those of method A. The classifier can be trained on both aerial data set and satellite data set to a nearly identical classification capability. The effects we observed during cross testing of classifiers using method A also hold for method B: the classifier trained on satellite imagery is better at classifying aerial imagery than the classifier trained on aerial imagery is at classifying satellite imagery. All the combinations of classifiers and data sets are capable of finding roads and since our cost map generator is built to deal with uncertainties of classifier we can use them for at least preliminary path planning. The conclusions from comparing the satellite data to aerial imagery: Correctly classified: 89,9% Houses, rate=97,37% Roads, rate=96,93% Grass, rate=96,33% Debris, rate=98,63% point for re-training on aerial imagery. Reverse also holds true; the classifier that has been trained on aerial imagery is a god starting point for re-training on satellite imagery. When using a trained classifier on a different imagery we must take into account the changes in nature - where there is grass on summer there is mud on spring and snow in winter. Each has different color and texture. Because the roads are easily distinguishable it is better to use non-matching classifier on terrain where roads are present than not to use overhead imagery at all. 3. INTEGRATION OF COST MAP GENERATION AND PATH PLANNING

6 The cost map is a grid on nodes; each node has a cost associated with it. For generating the cost map we divide an aerial photo into a grid of 29x29 pixel patterns and feed the patterns to our classifier which outputs feature vectors for the patterns. The pattern size is determined by the size of classifier input layer; to enhance the cost map resolution we use overlapping patterns. The feature vectors are mapped into a probability vectors using a threshold vector. For cost calculation the probability vectors are multiplied with a weight vector and a constant bias is added [4] [5]. cost = d f i, t i w i + c u, i where F is classifier output vector, T is threshold vector, d is the transformation function that maps classifier output to probability [5], W is the weight vector, c u is constant bias. It is important to note here that for calculating the cost of each node we need to evaluate the pattern associated with the node using a classifier. The evaluation of the pattern is relatively expensive operation when compared to the cost calculation. For path planning we use A* algorithm [6] which is improvement over Dijkstra's algorithm [7] and generates an optimal path from start node to end node. There s an excellent tutorial on A* algorithm by Patrick Lester: 1) Add the starting square (or node) to the open list. 2) Repeat the following: a) Look for the lowest F cost square on the open list. We refer to this as the current square. b) Switch it to the closed list. c) For each of the 8 squares adjacent to this current square If it /.../ is on the closed list, ignore it. Otherwise do the following. If it isn t on the open list, add it to the open list. Make the current square the parent of this square. Record the F, G, and H costs of the square. If it is on the open list already, check to see if this path to that square is better, using G cost as the measure. A lower G cost means that this is a better path. If so, change the parent of the square to the current square, and recalculate the G and F scores of the square. /.../ d) Stop when you: Add the target square to the closed list, in which case the path has been found /.../, or Fail to find the target square, and the open list is empty. In this case, there is no path. 3) Save the path. Working backwards from the target square, go from each square to its parent square until you reach the starting square. That is your path. [8] H is an heuristic value that estimates the cost of moving from current node to end square, G is the cost of moving from start node to current node and F = G + H. When calculating the cost G for any given neighboring node we add the cost G of current node to value taken from cost map. Because the A* algorithm does not necessarily use all nodes in cost map and because the generation of whole cost map is relatively slow process (when compared to path planning stage) we can get significant improvements to our algorithm by combining the path planner and the cost map generation steps. Illustration 4: red line presents the used path and the nodes evaluated by path planner are marked using dark area. Overall there was need to evaluate 7% of the aerial image for the path planning When we calculated the cost of a node only when it was needed by path planner we can occasionally reduce the combined time of cost map generation and path planning by

7 order of magnitude - especially when the path from start to end point is clear (illustration 1, illustration 2). However, this optimization technique does not affect the worst case scenario - there are cases where we still need to evaluate the whole imagery for path planning particularly in scenarios where there are no clear pathways to designated target position. Illustration 5: red line presents the used path and the nodes evaluated by path planner are marked using dark area. Overall there was need to evaluate 25% of the aerial image for the path planning. 4. CONCLUSIONS Our proposed system of extending UGV perception distance by extending UGV local experiences to wider area using aerial imagery works equally well using satellite imagery. We can also use classifier that has been trained for satellite imagery on aerial imagery and using of a classifier that is trained on similar imagery is a good starting point for re-training it for specific imagery in terms of quick results. By combining the cost map generation and path planning stages we can reduce the total time it takes for cost map generation and path planning up to 10x. The performance increase comes from executing the classifier on overhead imagery (expensive operation) as needed by path planner. The optimization, however, does not affect the worst case scenario where we still need to generate whole cost map. 5. BIBLIOGRAPHY [1] M. Hiiemaa and M. Tamre, "Semiautonomous Motion Control Layer for UGV-Type Robot," pp , [2] R. Hudjakov and M. Tamre, "Aerial Imagery Terrain Classification for Long-Range Autonomous Navigation," Proc. of International Symposium on Optomechatronic Technologies, pp , [3] R. Hudjakov and M. Tamre, "Aerial Imagery Terrain Classification for Long-Range Autonomous Navigation," Proc. of the 7th International Conference of DAAAM Baltic Industrial Engineering, pp , [4] R. Hudjakov and M. Tamre, "Ortophoto Analysis for UGV Long-Range Autonomous Navigation," Estonian Journal of Engineering, no. 17, p , [5] R. Hudjakov and M. Tamre, "Aerial Imagery Based Long-Range Path Planning for Unmanned Ground Vehicle," in Mechatronic Systems and Materials, Kaunas, [6] P. E. Hart, N. J. Nilsson and B. Raphael, "Correction to A Formal Basis for the Heuristic Determination of Minimum Cost Paths," SIGART Bull., no. 37, pp , December [7] E. W. Dijkstra, "A note on two problems in connexion with graphs," Numerische Mathematik, vol. 1, no. 1, pp , [8] P. Lester, "A* Pathfinding for Beginners," 18 July [Online]. Available: StarTutorial.htm. [Accessed ].

Basic Motion Planning Algorithms

Basic Motion Planning Algorithms Basic Motion Planning Algorithms Sohaib Younis Intelligent Robotics 7 January, 2013 1 Outline Introduction Environment Map Dijkstra's algorithm A* algorithm Summary 2 Introduction Definition: Motion Planning

More information

UNIVERSITY OF NORTH CAROLINA AT CHARLOTTE

UNIVERSITY OF NORTH CAROLINA AT CHARLOTTE UNIVERSITY OF NORTH CAROLINA AT CHARLOTTE Department of Electrical and Computer Engineering ECGR 4161/5196 Introduction to Robotics Experiment No. 5 A* Path Planning Overview: The purpose of this experiment

More information

Terrain Roughness Identification for High-Speed UGVs

Terrain Roughness Identification for High-Speed UGVs Proceedings of the International Conference of Control, Dynamic Systems, and Robotics Ottawa, Ontario, Canada, May 15-16 2014 Paper No. 11 Terrain Roughness Identification for High-Speed UGVs Graeme N.

More information

Autonomous Navigation in Unknown Environments via Language Grounding

Autonomous Navigation in Unknown Environments via Language Grounding Autonomous Navigation in Unknown Environments via Language Grounding Koushik (kbhavani) Aditya (avmandal) Sanjay (svnaraya) Mentor Jean Oh Introduction As robots become an integral part of various domains

More information

Three-Dimensional Off-Line Path Planning for Unmanned Aerial Vehicle Using Modified Particle Swarm Optimization

Three-Dimensional Off-Line Path Planning for Unmanned Aerial Vehicle Using Modified Particle Swarm Optimization Three-Dimensional Off-Line Path Planning for Unmanned Aerial Vehicle Using Modified Particle Swarm Optimization Lana Dalawr Jalal Abstract This paper addresses the problem of offline path planning for

More information

CONTROL OF AUTONOMOUS MOBILE ROBOTS USING WIRELESS NETWORK AND MATLAB

CONTROL OF AUTONOMOUS MOBILE ROBOTS USING WIRELESS NETWORK AND MATLAB 11th International DAAAM Baltic Conference INDUSTRIAL ENGINEERING 20-22 nd April 2016, Tallinn, Estonia CONTROL OF AUTONOMOUS MOBILE ROBOTS USING WIRELESS NETWORK AND MATLAB Vu TrieuMinh; Tamre M; Hudjakov

More information

Map Abstraction with Adjustable Time Bounds

Map Abstraction with Adjustable Time Bounds Map Abstraction with Adjustable Time Bounds Sourodeep Bhattacharjee and Scott D. Goodwin School of Computer Science, University of Windsor Windsor, N9B 3P4, Canada sourodeepbhattacharjee@gmail.com, sgoodwin@uwindsor.ca

More information

Rectification Algorithm for Linear Pushbroom Image of UAV

Rectification Algorithm for Linear Pushbroom Image of UAV Rectification Algorithm for Linear Pushbroom Image of UAV Ruoming SHI and Ling ZHU INTRODUCTION In recent years, unmanned aerial vehicle (UAV) has become a strong supplement and an important complement

More information

Mini Survey Paper (Robotic Mapping) Ryan Hamor CPRE 583 September 2011

Mini Survey Paper (Robotic Mapping) Ryan Hamor CPRE 583 September 2011 Mini Survey Paper (Robotic Mapping) Ryan Hamor CPRE 583 September 2011 Introduction The goal of this survey paper is to examine the field of robotic mapping and the use of FPGAs in various implementations.

More information

Autonomous robot motion path planning using shortest path planning algorithms

Autonomous robot motion path planning using shortest path planning algorithms IOSR Journal of Engineering (IOSRJEN) e-issn: 2250-3021, p-issn: 2278-8719 Vol. 3, Issue 1 (Jan. 2013), V1 PP 65-69 Autonomous robot motion path planning using shortest path planning algorithms T. Ravi

More information

AN IMPLEMENTATION OF PATH PLANNING ALGORITHMS FOR MOBILE ROBOTS ON A GRID BASED MAP

AN IMPLEMENTATION OF PATH PLANNING ALGORITHMS FOR MOBILE ROBOTS ON A GRID BASED MAP AN IMPLEMENTATION OF PATH PLANNING ALGORITHMS FOR MOBILE ROBOTS ON A GRID BASED MAP Tolga YÜKSEL Abdullah SEZGİN e-mail : tyuksel@omu.edu.tr e-mail : asezgin@omu.edu.tr Ondokuz Mayıs University, Electrical

More information

Homework #6 (Constraint Satisfaction, Non-Deterministic Uncertainty and Adversarial Search) Out: 2/21/11 Due: 2/29/11 (at noon)

Homework #6 (Constraint Satisfaction, Non-Deterministic Uncertainty and Adversarial Search) Out: 2/21/11 Due: 2/29/11 (at noon) CS121 Introduction to Artificial Intelligence Winter 2011 Homework #6 (Constraint Satisfaction, Non-Deterministic Uncertainty and Adversarial Search) Out: 2/21/11 Due: 2/29/11 (at noon) How to complete

More information

A System for Bidirectional Robotic Pathfinding

A System for Bidirectional Robotic Pathfinding A System for Bidirectional Robotic Pathfinding Tesca K. Fitzgerald Department of Computer Science, Portland State University PO Box 751 Portland, OR 97207 USA tesca@cs.pdx.edu TR 12-02 November 2012 Abstract

More information

Pedestrian Detection Using Correlated Lidar and Image Data EECS442 Final Project Fall 2016

Pedestrian Detection Using Correlated Lidar and Image Data EECS442 Final Project Fall 2016 edestrian Detection Using Correlated Lidar and Image Data EECS442 Final roject Fall 2016 Samuel Rohrer University of Michigan rohrer@umich.edu Ian Lin University of Michigan tiannis@umich.edu Abstract

More information

Pathfinding in partially explored games environments

Pathfinding in partially explored games environments Pathfinding in partially explored games environments The application of the A Algorithm with Occupancy Grids in Unity3D John Stamford, Arjab Singh Khuman, Jenny Carter & Samad Ahmadi Centre for Computational

More information

AUTONOMOUS IMAGE EXTRACTION AND SEGMENTATION OF IMAGE USING UAV S

AUTONOMOUS IMAGE EXTRACTION AND SEGMENTATION OF IMAGE USING UAV S AUTONOMOUS IMAGE EXTRACTION AND SEGMENTATION OF IMAGE USING UAV S Radha Krishna Rambola, Associate Professor, NMIMS University, India Akash Agrawal, Student at NMIMS University, India ABSTRACT Due to the

More information

Automated Route Finding on Digital Terrains

Automated Route Finding on Digital Terrains Automated Route Finding on Digital Terrains D. R. Wichmann and B. C. Wünsche Graphics Group, Dept. of Computer Science, University of Auckland, Private Bag 92019, Auckland, New Zealand. daniel.wichmann@gmail.com,

More information

Beyond A*: Speeding up pathfinding through hierarchical abstraction. Daniel Harabor NICTA & The Australian National University

Beyond A*: Speeding up pathfinding through hierarchical abstraction. Daniel Harabor NICTA & The Australian National University Beyond A*: Speeding up pathfinding through hierarchical abstraction Daniel Harabor NICTA & The Australian National University Motivation Myth: Pathfinding is a solved problem. Motivation Myth: Pathfinding

More information

INTEGRATING LOCAL AND GLOBAL NAVIGATION IN UNMANNED GROUND VEHICLES

INTEGRATING LOCAL AND GLOBAL NAVIGATION IN UNMANNED GROUND VEHICLES INTEGRATING LOCAL AND GLOBAL NAVIGATION IN UNMANNED GROUND VEHICLES Juan Pablo Gonzalez*, William Dodson, Robert Dean General Dynamics Robotic Systems Westminster, MD Alberto Lacaze, Leonid Sapronov Robotics

More information

Domain Adaptation For Mobile Robot Navigation

Domain Adaptation For Mobile Robot Navigation Domain Adaptation For Mobile Robot Navigation David M. Bradley, J. Andrew Bagnell Robotics Institute Carnegie Mellon University Pittsburgh, 15217 dbradley, dbagnell@rec.ri.cmu.edu 1 Introduction An important

More information

Artificial Neural Network-Based Prediction of Human Posture

Artificial Neural Network-Based Prediction of Human Posture Artificial Neural Network-Based Prediction of Human Posture Abstract The use of an artificial neural network (ANN) in many practical complicated problems encourages its implementation in the digital human

More information

THREE DIMENSIONAL CURVE HALL RECONSTRUCTION USING SEMI-AUTOMATIC UAV

THREE DIMENSIONAL CURVE HALL RECONSTRUCTION USING SEMI-AUTOMATIC UAV THREE DIMENSIONAL CURVE HALL RECONSTRUCTION USING SEMI-AUTOMATIC UAV Muhammad Norazam Zulgafli 1 and Khairul Nizam Tahar 1,2 1 Centre of Studies for Surveying Science and Geomatics, Faculty of Architecture

More information

Lab 9. Julia Janicki. Introduction

Lab 9. Julia Janicki. Introduction Lab 9 Julia Janicki Introduction My goal for this project is to map a general land cover in the area of Alexandria in Egypt using supervised classification, specifically the Maximum Likelihood and Support

More information

Practice Exam Sample Solutions

Practice Exam Sample Solutions CS 675 Computer Vision Instructor: Marc Pomplun Practice Exam Sample Solutions Note that in the actual exam, no calculators, no books, and no notes allowed. Question 1: out of points Question 2: out of

More information

Jo-Car2 Autonomous Mode. Path Planning (Cost Matrix Algorithm)

Jo-Car2 Autonomous Mode. Path Planning (Cost Matrix Algorithm) Chapter 8.2 Jo-Car2 Autonomous Mode Path Planning (Cost Matrix Algorithm) Introduction: In order to achieve its mission and reach the GPS goal safely; without crashing into obstacles or leaving the lane,

More information

Anytime, Dynamic Planning in High-dimensional Search Spaces

Anytime, Dynamic Planning in High-dimensional Search Spaces 2007 IEEE International Conference on Robotics and Automation Roma, Italy, 10-14 April 2007 WeD11.2 Anytime, Dynamic Planning in High-dimensional Search Spaces Dave Ferguson Intel Research Pittsburgh 4720

More information

Wave front Method Based Path Planning Algorithm for Mobile Robots

Wave front Method Based Path Planning Algorithm for Mobile Robots Wave front Method Based Path Planning Algorithm for Mobile Robots Bhavya Ghai 1 and Anupam Shukla 2 ABV- Indian Institute of Information Technology and Management, Gwalior, India 1 bhavyaghai@gmail.com,

More information

Mapping Road surface condition using Unmanned Aerial Vehicle- Based Imaging System. Ahmed F. Elaksher St. Cloud State University

Mapping Road surface condition using Unmanned Aerial Vehicle- Based Imaging System. Ahmed F. Elaksher St. Cloud State University Mapping Road surface condition using Unmanned Aerial Vehicle- Based Imaging System Ahmed F. Elaksher St. Cloud State University 1 Outline Introduction & Motivation Methodology Experimental Results & Analysis

More information

Opening the Black Box Data Driven Visualizaion of Neural N

Opening the Black Box Data Driven Visualizaion of Neural N Opening the Black Box Data Driven Visualizaion of Neural Networks September 20, 2006 Aritificial Neural Networks Limitations of ANNs Use of Visualization (ANNs) mimic the processes found in biological

More information

Kaijen Hsiao. Part A: Topics of Fascination

Kaijen Hsiao. Part A: Topics of Fascination Kaijen Hsiao Part A: Topics of Fascination 1) I am primarily interested in SLAM. I plan to do my project on an application of SLAM, and thus I am interested not only in the method we learned about in class,

More information

Lesson 1 Introduction to Path Planning Graph Searches: BFS and DFS

Lesson 1 Introduction to Path Planning Graph Searches: BFS and DFS Lesson 1 Introduction to Path Planning Graph Searches: BFS and DFS DASL Summer Program Path Planning References: http://robotics.mem.drexel.edu/mhsieh/courses/mem380i/index.html http://dasl.mem.drexel.edu/hing/bfsdfstutorial.htm

More information

Searching with Partial Information

Searching with Partial Information Searching with Partial Information Above we (unrealistically) assumed that the environment is fully observable and deterministic Moreover, we assumed that the agent knows what the effects of each action

More information

3D LIDAR Point Cloud based Intersection Recognition for Autonomous Driving

3D LIDAR Point Cloud based Intersection Recognition for Autonomous Driving 3D LIDAR Point Cloud based Intersection Recognition for Autonomous Driving Quanwen Zhu, Long Chen, Qingquan Li, Ming Li, Andreas Nüchter and Jian Wang Abstract Finding road intersections in advance is

More information

Discrete search algorithms

Discrete search algorithms Robot Autonomy (16-662, S13) Lecture#08 (Monday February 11) Discrete search algorithms Lecturer: Siddhartha Srinivasa Scribes: Kavya Suresh & David Matten I. INTRODUCTION These notes contain a detailed

More information

LECTURE NOTES Professor Anita Wasilewska NEURAL NETWORKS

LECTURE NOTES Professor Anita Wasilewska NEURAL NETWORKS LECTURE NOTES Professor Anita Wasilewska NEURAL NETWORKS Neural Networks Classifier Introduction INPUT: classification data, i.e. it contains an classification (class) attribute. WE also say that the class

More information

Unmanned Vehicle Technology Researches for Outdoor Environments. *Ju-Jang Lee 1)

Unmanned Vehicle Technology Researches for Outdoor Environments. *Ju-Jang Lee 1) Keynote Paper Unmanned Vehicle Technology Researches for Outdoor Environments *Ju-Jang Lee 1) 1) Department of Electrical Engineering, KAIST, Daejeon 305-701, Korea 1) jjlee@ee.kaist.ac.kr ABSTRACT The

More information

6.034 Quiz 2, Spring 2005

6.034 Quiz 2, Spring 2005 6.034 Quiz 2, Spring 2005 Open Book, Open Notes Name: Problem 1 (13 pts) 2 (8 pts) 3 (7 pts) 4 (9 pts) 5 (8 pts) 6 (16 pts) 7 (15 pts) 8 (12 pts) 9 (12 pts) Total (100 pts) Score 1 1 Decision Trees (13

More information

Map-Based Strategies for Robot Navigation in Unknown Environments

Map-Based Strategies for Robot Navigation in Unknown Environments Map-Based trategies for Robot Navigation in Unknown Environments Anthony tentz Robotics Institute Carnegie Mellon University Pittsburgh, Pennsylvania 15213 U.. A. Abstract Robot path planning algorithms

More information

Technical Paper of HITCSC Team for The International Aerial Robotics Competition

Technical Paper of HITCSC Team for The International Aerial Robotics Competition Technical Paper of HITCSC Team for The International Aerial Robotics Competition Qingtao Yu, Yuanhong Li, Hang Yu, Haodi Yao, Rui Xing, Qi Zhao, Xiaowei Xing Harbin Institute of Tecnology ABSTRACT This

More information

Improvement of Automated Guided Vehicle s image recognition - Object detection and identification Zhu Xin

Improvement of Automated Guided Vehicle s image recognition - Object detection and identification Zhu Xin DEGREE PROJECT FOR MASTER OF SCIENCE WITH SPECIALIZATION IN ROBOTICS DEPARTMENT OF ENGINEERING SCIENCE UNIVERSITY WEST Improvement of Automated Guided Vehicle s image recognition - Object detection and

More information

Reality Modeling Drone Capture Guide

Reality Modeling Drone Capture Guide Reality Modeling Drone Capture Guide Discover the best practices for photo acquisition-leveraging drones to create 3D reality models with ContextCapture, Bentley s reality modeling software. Learn the

More information

Spring Localization II. Roland Siegwart, Margarita Chli, Martin Rufli. ASL Autonomous Systems Lab. Autonomous Mobile Robots

Spring Localization II. Roland Siegwart, Margarita Chli, Martin Rufli. ASL Autonomous Systems Lab. Autonomous Mobile Robots Spring 2016 Localization II Localization I 25.04.2016 1 knowledge, data base mission commands Localization Map Building environment model local map position global map Cognition Path Planning path Perception

More information

A MAPPING SIMULATOR. Keywords: mapping simulator, path finding, irobot Roomba, OpenGL

A MAPPING SIMULATOR. Keywords: mapping simulator, path finding, irobot Roomba, OpenGL A MAPPING SIMULATOR Catalin-Antonio Dedu 1 Costin-Anton Boiangiu 2 Ion Bucur 3 ABSTRACT This paper approaches the domain of mapping simulations and presents the result of subsequent research in path finding

More information

CS 4758 Robot Navigation Through Exit Sign Detection

CS 4758 Robot Navigation Through Exit Sign Detection CS 4758 Robot Navigation Through Exit Sign Detection Aaron Sarna Michael Oleske Andrew Hoelscher Abstract We designed a set of algorithms that utilize the existing corridor navigation code initially created

More information

COMPARISION OF REGRESSION WITH NEURAL NETWORK MODEL FOR THE VARIATION OF VANISHING POINT WITH VIEW ANGLE IN DEPTH ESTIMATION WITH VARYING BRIGHTNESS

COMPARISION OF REGRESSION WITH NEURAL NETWORK MODEL FOR THE VARIATION OF VANISHING POINT WITH VIEW ANGLE IN DEPTH ESTIMATION WITH VARYING BRIGHTNESS International Journal of Advanced Trends in Computer Science and Engineering, Vol.2, No.1, Pages : 171-177 (2013) COMPARISION OF REGRESSION WITH NEURAL NETWORK MODEL FOR THE VARIATION OF VANISHING POINT

More information

COS Lecture 13 Autonomous Robot Navigation

COS Lecture 13 Autonomous Robot Navigation COS 495 - Lecture 13 Autonomous Robot Navigation Instructor: Chris Clark Semester: Fall 2011 1 Figures courtesy of Siegwart & Nourbakhsh Control Structure Prior Knowledge Operator Commands Localization

More information

Efficient Route Finding and Sensors for Collision Detection in Google s Driverless Car

Efficient Route Finding and Sensors for Collision Detection in Google s Driverless Car Available Online at www.ijcsmc.com International Journal of Computer Science and Mobile Computing A Monthly Journal of Computer Science and Information Technology IJCSMC, Vol. 3, Issue. 12, December 2014,

More information

Spring Localization II. Roland Siegwart, Margarita Chli, Juan Nieto, Nick Lawrance. ASL Autonomous Systems Lab. Autonomous Mobile Robots

Spring Localization II. Roland Siegwart, Margarita Chli, Juan Nieto, Nick Lawrance. ASL Autonomous Systems Lab. Autonomous Mobile Robots Spring 2018 Localization II Localization I 16.04.2018 1 knowledge, data base mission commands Localization Map Building environment model local map position global map Cognition Path Planning path Perception

More information

Fast Incremental Learning for Off-Road Robot Navigation

Fast Incremental Learning for Off-Road Robot Navigation Fast Incremental Learning for Off-Road Robot Navigation Beat Flepp NVIDIA Corp Holmdel NJ 07733 Artem Provodin New York University New York NY 10003 L. D. Jackel North C Technologies Holmdel NJ 07733 Yann

More information

Chapter 10 Motion Planning

Chapter 10 Motion Planning Chapter 10 Motion Planning Part 3 10.3 Real Time Global Motion Planning 1 Outline 10.3 Real Time Global Motion Planning 10.3.1 Introduction 10.3.2 Depth Limited Approaches 10.3.3 Anytime Approaches 10.3.4

More information

Path Planning with Dynamic Obstacle Avoidance for a Jumping-Enabled Robot

Path Planning with Dynamic Obstacle Avoidance for a Jumping-Enabled Robot Path Planning with Dynamic Obstacle Avoidance for a Jumping-Enabled Robot Peggy (Yuchun) Wang and Caitlin Hogan Department of Computer Science, Stanford University Robotic navigation is an important concern

More information

3D Convolutional Neural Networks for Landing Zone Detection from LiDAR

3D Convolutional Neural Networks for Landing Zone Detection from LiDAR 3D Convolutional Neural Networks for Landing Zone Detection from LiDAR Daniel Mataruna and Sebastian Scherer Presented by: Sabin Kafle Outline Introduction Preliminaries Approach Volumetric Density Mapping

More information

GPS Located Accuracy Assessment Plots on the Modoc National Forest

GPS Located Accuracy Assessment Plots on the Modoc National Forest This file was created by scanning the printed publication. Errors identified by the software have been corrected; however, some errors may remain. GPS Located Accuracy Assessment Plots on the Modoc National

More information

CS231A Course Project Final Report Sign Language Recognition with Unsupervised Feature Learning

CS231A Course Project Final Report Sign Language Recognition with Unsupervised Feature Learning CS231A Course Project Final Report Sign Language Recognition with Unsupervised Feature Learning Justin Chen Stanford University justinkchen@stanford.edu Abstract This paper focuses on experimenting with

More information

CONTROL ALGORITHMS FOR SPACE TUG RENDEZVOUS. Statement of Project Fall Author: Timothée de Mierry

CONTROL ALGORITHMS FOR SPACE TUG RENDEZVOUS. Statement of Project Fall Author: Timothée de Mierry CONTROL ALGORITHMS FOR SPACE TUG RENDEZVOUS Statement of Project 16.621 Fall 2002 Author: Timothée de Mierry Advisor: Olivier de Weck Partner: Gergana Bounova Thursday, October 3 rd 1. Introduction The

More information

Bidirectional search and Goal-directed Dijkstra

Bidirectional search and Goal-directed Dijkstra Computing the shortest path Bidirectional search and Goal-directed Dijkstra Alexander Kozyntsev October 18, 2010 Abstract We study the problem of finding a shortest path between two vertices in a directed

More information

Attack Resilient State Estimation for Vehicular Systems

Attack Resilient State Estimation for Vehicular Systems December 15 th 2013. T-SET Final Report Attack Resilient State Estimation for Vehicular Systems Nicola Bezzo (nicbezzo@seas.upenn.edu) Prof. Insup Lee (lee@cis.upenn.edu) PRECISE Center University of Pennsylvania

More information

APPLICATION OF AERIAL VIDEO FOR TRAFFIC FLOW MONITORING AND MANAGEMENT

APPLICATION OF AERIAL VIDEO FOR TRAFFIC FLOW MONITORING AND MANAGEMENT Pitu Mirchandani, Professor, Department of Systems and Industrial Engineering Mark Hickman, Assistant Professor, Department of Civil Engineering Alejandro Angel, Graduate Researcher Dinesh Chandnani, Graduate

More information

Introduction to Information Science and Technology (IST) Part IV: Intelligent Machines and Robotics Planning

Introduction to Information Science and Technology (IST) Part IV: Intelligent Machines and Robotics Planning Introduction to Information Science and Technology (IST) Part IV: Intelligent Machines and Robotics Planning Sören Schwertfeger / 师泽仁 ShanghaiTech University ShanghaiTech University - SIST - 10.05.2017

More information

1 Lab + Hwk 5: Particle Swarm Optimization

1 Lab + Hwk 5: Particle Swarm Optimization 1 Lab + Hwk 5: Particle Swarm Optimization This laboratory requires the following equipment: C programming tools (gcc, make), already installed in GR B001 Webots simulation software Webots User Guide Webots

More information

A Deep Learning Approach to Vehicle Speed Estimation

A Deep Learning Approach to Vehicle Speed Estimation A Deep Learning Approach to Vehicle Speed Estimation Benjamin Penchas bpenchas@stanford.edu Tobin Bell tbell@stanford.edu Marco Monteiro marcorm@stanford.edu ABSTRACT Given car dashboard video footage,

More information

MOVING OBJECT DETECTION USING BACKGROUND SUBTRACTION ALGORITHM USING SIMULINK

MOVING OBJECT DETECTION USING BACKGROUND SUBTRACTION ALGORITHM USING SIMULINK MOVING OBJECT DETECTION USING BACKGROUND SUBTRACTION ALGORITHM USING SIMULINK Mahamuni P. D 1, R. P. Patil 2, H.S. Thakar 3 1 PG Student, E & TC Department, SKNCOE, Vadgaon Bk, Pune, India 2 Asst. Professor,

More information

CS 354R: Computer Game Technology

CS 354R: Computer Game Technology CS 354R: Computer Game Technology A* Heuristics Fall 2018 A* Search f(n): The current best estimate for the best path through a node: f(n)=g(n)+h(n) g(n): current known best cost for getting to a node

More information

Locality through Modular Network Topology

Locality through Modular Network Topology Locality through Modular Network Topology Oliver Lipton and Harry Huchra December 18, 2017 Abstract Our experiment concerns designing a robot to accomplish the task of locality grounding. Through a dataset

More information

Fast Local Planner for Autonomous Helicopter

Fast Local Planner for Autonomous Helicopter Fast Local Planner for Autonomous Helicopter Alexander Washburn talexan@seas.upenn.edu Faculty advisor: Maxim Likhachev April 22, 2008 Abstract: One challenge of autonomous flight is creating a system

More information

Learning Composite Operators for Object Detection

Learning Composite Operators for Object Detection Real-World Applications Learning Composite Operators for Object Detection Bir Bhanu and Yingqiang Lin Center for Research in Intelligent Systems University of California, Riverside, CA, 92521, USA Email:

More information

Terrain Data Real-time Analysis Based on Point Cloud for Mars Rover

Terrain Data Real-time Analysis Based on Point Cloud for Mars Rover Terrain Data Real-time Analysis Based on Point Cloud for Mars Rover Haoruo ZHANG 1, Yuanjie TAN, Qixin CAO. Abstract. With the development of space exploration, more and more aerospace researchers pay

More information

Remote Sensing Image Analysis via a Texture Classification Neural Network

Remote Sensing Image Analysis via a Texture Classification Neural Network Remote Sensing Image Analysis via a Texture Classification Neural Network Hayit K. Greenspan and Rodney Goodman Department of Electrical Engineering California Institute of Technology, 116-81 Pasadena,

More information

APPLICATION OF SOFTMAX REGRESSION AND ITS VALIDATION FOR SPECTRAL-BASED LAND COVER MAPPING

APPLICATION OF SOFTMAX REGRESSION AND ITS VALIDATION FOR SPECTRAL-BASED LAND COVER MAPPING APPLICATION OF SOFTMAX REGRESSION AND ITS VALIDATION FOR SPECTRAL-BASED LAND COVER MAPPING J. Wolfe a, X. Jin a, T. Bahr b, N. Holzer b, * a Harris Corporation, Broomfield, Colorado, U.S.A. (jwolfe05,

More information

1 Lab 5: Particle Swarm Optimization

1 Lab 5: Particle Swarm Optimization 1 Lab 5: Particle Swarm Optimization This laboratory requires the following: (The development tools are installed in GR B0 01 already): C development tools (gcc, make, etc.) Webots simulation software

More information

Deep Convolutional Neural Networks. Nov. 20th, 2015 Bruce Draper

Deep Convolutional Neural Networks. Nov. 20th, 2015 Bruce Draper Deep Convolutional Neural Networks Nov. 20th, 2015 Bruce Draper Background: Fully-connected single layer neural networks Feed-forward classification Trained through back-propagation Example Computer Vision

More information

Autonomous Robot Navigation: Using Multiple Semi-supervised Models for Obstacle Detection

Autonomous Robot Navigation: Using Multiple Semi-supervised Models for Obstacle Detection Autonomous Robot Navigation: Using Multiple Semi-supervised Models for Obstacle Detection Adam Bates University of Colorado at Boulder Abstract: This paper proposes a novel approach to efficiently creating

More information

Robust line segmentation for handwritten documents

Robust line segmentation for handwritten documents Robust line segmentation for handwritten documents Kamal Kuzhinjedathu, Harish Srinivasan and Sargur Srihari Center of Excellence for Document Analysis and Recognition (CEDAR) University at Buffalo, State

More information

6. NEURAL NETWORK BASED PATH PLANNING ALGORITHM 6.1 INTRODUCTION

6. NEURAL NETWORK BASED PATH PLANNING ALGORITHM 6.1 INTRODUCTION 6 NEURAL NETWORK BASED PATH PLANNING ALGORITHM 61 INTRODUCTION In previous chapters path planning algorithms such as trigonometry based path planning algorithm and direction based path planning algorithm

More information

Application of Deep Learning Techniques in Satellite Telemetry Analysis.

Application of Deep Learning Techniques in Satellite Telemetry Analysis. Application of Deep Learning Techniques in Satellite Telemetry Analysis. Greg Adamski, Member of Technical Staff L3 Technologies Telemetry and RF Products Julian Spencer Jones, Spacecraft Engineer Telenor

More information

Geometric Path Planning for General Robot Manipulators

Geometric Path Planning for General Robot Manipulators Proceedings of the World Congress on Engineering and Computer Science 29 Vol II WCECS 29, October 2-22, 29, San Francisco, USA Geometric Path Planning for General Robot Manipulators Ziyad Aljarboua Abstract

More information

Lecture 3: Linear Classification

Lecture 3: Linear Classification Lecture 3: Linear Classification Roger Grosse 1 Introduction Last week, we saw an example of a learning task called regression. There, the goal was to predict a scalar-valued target from a set of features.

More information

Artifacts and Textured Region Detection

Artifacts and Textured Region Detection Artifacts and Textured Region Detection 1 Vishal Bangard ECE 738 - Spring 2003 I. INTRODUCTION A lot of transformations, when applied to images, lead to the development of various artifacts in them. In

More information

HARNESSING CERTAINTY TO SPEED TASK-ALLOCATION ALGORITHMS FOR MULTI-ROBOT SYSTEMS

HARNESSING CERTAINTY TO SPEED TASK-ALLOCATION ALGORITHMS FOR MULTI-ROBOT SYSTEMS HARNESSING CERTAINTY TO SPEED TASK-ALLOCATION ALGORITHMS FOR MULTI-ROBOT SYSTEMS An Undergraduate Research Scholars Thesis by DENISE IRVIN Submitted to the Undergraduate Research Scholars program at Texas

More information

Neural Networks. CE-725: Statistical Pattern Recognition Sharif University of Technology Spring Soleymani

Neural Networks. CE-725: Statistical Pattern Recognition Sharif University of Technology Spring Soleymani Neural Networks CE-725: Statistical Pattern Recognition Sharif University of Technology Spring 2013 Soleymani Outline Biological and artificial neural networks Feed-forward neural networks Single layer

More information

Neural Networks Terrain Classification using Inertial Measurement Unit for an Autonomous Vehicle

Neural Networks Terrain Classification using Inertial Measurement Unit for an Autonomous Vehicle SICE Annual Conference 2008 August 20-22, 2008, The University Electro-Communications, Japan Neural Networks Terrain Classification using Inertial Measurement Unit for an Autonomous Vehicle Rukwan Jitpakdee

More information

ESTIMATING THE COST OF ENERGY USAGE IN SPORT CENTRES: A COMPARATIVE MODELLING APPROACH

ESTIMATING THE COST OF ENERGY USAGE IN SPORT CENTRES: A COMPARATIVE MODELLING APPROACH ESTIMATING THE COST OF ENERGY USAGE IN SPORT CENTRES: A COMPARATIVE MODELLING APPROACH A.H. Boussabaine, R.J. Kirkham and R.G. Grew Construction Cost Engineering Research Group, School of Architecture

More information

Learning Image-Based Landmarks for Wayfinding using Neural Networks

Learning Image-Based Landmarks for Wayfinding using Neural Networks Learning Image-Based Landmarks for Wayfinding using Neural Networks Jeremy D. Drysdale Damian M. Lyons Robotics and Computer Vision Laboratory Department of Computer Science 340 JMH Fordham University

More information

2. POINT CLOUD DATA PROCESSING

2. POINT CLOUD DATA PROCESSING Point Cloud Generation from suas-mounted iphone Imagery: Performance Analysis A. D. Ladai, J. Miller Towill, Inc., 2300 Clayton Road, Suite 1200, Concord, CA 94520-2176, USA - (andras.ladai, jeffrey.miller)@towill.com

More information

A Hierarchical Approach to Probabilistic Pursuit-Evasion Games with Unmanned Ground and Aerial Vehicles 1

A Hierarchical Approach to Probabilistic Pursuit-Evasion Games with Unmanned Ground and Aerial Vehicles 1 A Hierarchical Approach to Probabilistic Pursuit-Evasion Games with Unmanned Ground and Aerial Vehicles 1 H. Jin Kim René Vidal David H. Shim Omid Shakernia Shankar Sastry Department of Electrical Engineering

More information

Harnessing GIS and Imagery for Power Transmission Inspection. ESRI European Users Conference October 15, 2015

Harnessing GIS and Imagery for Power Transmission Inspection. ESRI European Users Conference October 15, 2015 Harnessing GIS and Imagery for Power Transmission Inspection ESRI European Users Conference October 15, 2015 About Us Airborne/Threod Designer, manufacturer & service provider for multi-rotor & fixed wing

More information

OBJECT detection in general has many applications

OBJECT detection in general has many applications 1 Implementing Rectangle Detection using Windowed Hough Transform Akhil Singh, Music Engineering, University of Miami Abstract This paper implements Jung and Schramm s method to use Hough Transform for

More information

Face Recognition Using Vector Quantization Histogram and Support Vector Machine Classifier Rong-sheng LI, Fei-fei LEE *, Yan YAN and Qiu CHEN

Face Recognition Using Vector Quantization Histogram and Support Vector Machine Classifier Rong-sheng LI, Fei-fei LEE *, Yan YAN and Qiu CHEN 2016 International Conference on Artificial Intelligence: Techniques and Applications (AITA 2016) ISBN: 978-1-60595-389-2 Face Recognition Using Vector Quantization Histogram and Support Vector Machine

More information

Creating Affordable and Reliable Autonomous Vehicle Systems

Creating Affordable and Reliable Autonomous Vehicle Systems Creating Affordable and Reliable Autonomous Vehicle Systems Shaoshan Liu shaoshan.liu@perceptin.io Autonomous Driving Localization Most crucial task of autonomous driving Solutions: GNSS but withvariations,

More information

CS4758: Moving Person Avoider

CS4758: Moving Person Avoider CS4758: Moving Person Avoider Yi Heng Lee, Sze Kiat Sim Abstract We attempt to have a quadrotor autonomously avoid people while moving through an indoor environment. Our algorithm for detecting people

More information

Framed-Quadtree Path Planning for Mobile Robots Operating in Sparse Environments

Framed-Quadtree Path Planning for Mobile Robots Operating in Sparse Environments In Proceedings, IEEE Conference on Robotics and Automation, (ICRA), Leuven, Belgium, May 1998. Framed-Quadtree Path Planning for Mobile Robots Operating in Sparse Environments Alex Yahja, Anthony Stentz,

More information

Jane Li. Assistant Professor Mechanical Engineering Department, Robotic Engineering Program Worcester Polytechnic Institute

Jane Li. Assistant Professor Mechanical Engineering Department, Robotic Engineering Program Worcester Polytechnic Institute Jane Li Assistant Professor Mechanical Engineering Department, Robotic Engineering Program Worcester Polytechnic Institute A search-algorithm prioritizes and expands the nodes in its open list items by

More information

Character Recognition Using Convolutional Neural Networks

Character Recognition Using Convolutional Neural Networks Character Recognition Using Convolutional Neural Networks David Bouchain Seminar Statistical Learning Theory University of Ulm, Germany Institute for Neural Information Processing Winter 2006/2007 Abstract

More information

Face Detection. Gary Chern, Paul Gurney, and Jared Starman

Face Detection. Gary Chern, Paul Gurney, and Jared Starman Face Detection Gary Chern, Paul Gurney, and Jared Starman. Introduction Automatic face detection is a complex problem in image processing. Many methods exist to solve this problem such as template matching,

More information

Using Machine Learning for Classification of Cancer Cells

Using Machine Learning for Classification of Cancer Cells Using Machine Learning for Classification of Cancer Cells Camille Biscarrat University of California, Berkeley I Introduction Cell screening is a commonly used technique in the development of new drugs.

More information

Notes on Multilayer, Feedforward Neural Networks

Notes on Multilayer, Feedforward Neural Networks Notes on Multilayer, Feedforward Neural Networks CS425/528: Machine Learning Fall 2012 Prepared by: Lynne E. Parker [Material in these notes was gleaned from various sources, including E. Alpaydin s book

More information

Deep Learning for Visual Computing Prof. Debdoot Sheet Department of Electrical Engineering Indian Institute of Technology, Kharagpur

Deep Learning for Visual Computing Prof. Debdoot Sheet Department of Electrical Engineering Indian Institute of Technology, Kharagpur Deep Learning for Visual Computing Prof. Debdoot Sheet Department of Electrical Engineering Indian Institute of Technology, Kharagpur Lecture - 05 Classification with Perceptron Model So, welcome to today

More information

A motion planning method for mobile robot considering rotational motion in area coverage task

A motion planning method for mobile robot considering rotational motion in area coverage task Asia Pacific Conference on Robot IoT System Development and Platform 018 (APRIS018) A motion planning method for mobile robot considering rotational motion in area coverage task Yano Taiki 1,a) Takase

More information

Graph-based Planning Using Local Information for Unknown Outdoor Environments

Graph-based Planning Using Local Information for Unknown Outdoor Environments Graph-based Planning Using Local Information for Unknown Outdoor Environments Jinhan Lee, Roozbeh Mottaghi, Charles Pippin and Tucker Balch {jinhlee, roozbehm, cepippin, tucker}@cc.gatech.edu Center for

More information

A Comparison of Robot Navigation Algorithms for an Unknown Goal

A Comparison of Robot Navigation Algorithms for an Unknown Goal A Comparison of Robot Navigation Algorithms for an Unknown Goal Russell Bayuk Steven Ratering (faculty mentor) Computer Science Department University of Wisconsin Eau Claire Eau Claire, WI 54702 {bayukrj,

More information