9 th International AAAM Baltic Conference INUSTRIAL ENGINEERING - -6 April 01, Tallinn, Estonia AUTOMATION OF AN EXCAVATOR BOOM MOVEMENT IN -IMENSIONS Liukkonen, J.; Knuuttila, P.; Nguyen, T.; Ingale, S.; Kiviluoma, P.; Korkiala- Tanttu, L. & Kuosmanen, P. Abstract: Mass stabilization is the process of mixing binder into soil to improve its strength and stiffness. Stabilization is carried out using a specialized tool attached at the end of an excavator boom. The purpose of this project is to automate the boom movement of a mini excavator to better understand how automation affects the speed and accuracy of the system. The movement was automated using AT90CAN18 microcontroller connected by CAN bus to proportional valves which control the hydraulic cylinders of the boom based on the angle feedback from the resolvers. It was possible to move the bucket attached at the end of the boom reliably from point to point inside the reachable space of the excavator. Keywords: mass stabilization, robotic excavator, PI control, kinematics 1. INTROUCTION Mass stabilization is the process of hardening soft soil by mixing in a binding agent. Mass stabilization unit consists of one human operated excavator and a separate binding agent tank [ 1 ]. Typically large areas need to be stabilized and maintaining a consistent quality during this process is a difficult and time consuming task. The repetitive nature of the process suggests substantial benefits in automatizing the task. This research focuses on automating the boom movement. The purpose is to reduce the effect of control movements of the operator to achieve more accurate and efficient result of mass stabilization. Independent movement of the excavator together with the tank is left out of this research. The research builds on the previous studies carried out by Martikainen et al. [ ] and Kiviranta [ ]. The goal is to extend the results of one-dimensional movement of the bucket tip by Martikainen et al. into full three-dimensional movement. Other groups such as Lee et al. [ ] have also conducted research into automated boom movement for automated levelling work.. METHOS The excavator used in the project (JCB Micro 800) is automated using electrohydraulic valves (Sauer-anfoss, PVE-CC Series ) and resolvers (Axiomatic AXRES-CO-V-1X-H) which are connected through CAN bus to a microcontroller (AT90CAN18). The electrohydraulic valves of the excavator allow operation of the boom using a microcontroller and resolvers attached to each joint which provide accurate information of the joint angles. Since the original diesel engine of the excavator had been replaced with an electric motor, the resolvers were connected by twisted pair cables to minimize the electromagnetic interference. A printed circuit board and the supporting electronics were built to interface the microcontroller with the electrohydraulic valves and the resolvers. Bucket tip coordinates specified in the custom made 51
PC software were supplied to the microcontroller via serial port. In the microcontroller, a software PI controller adjusts the flow for each valve based on the difference between the reference angle and the current angle for each joint. The control messages for the valves are sent through CAN bus and define the flow and spool drive direction. Figure 1 has a flowchart of the interaction between the system components. θ = tan tan a p d oz + ( ) ( ) ) ( ) + d + d + a a ( ) + + d a a, () where p oz is coordinate of point O. Parameters a and a describe lengths of link and, and d used to shorten the equation is defined as d ( 1) pox + sin( 1) poy a1 = cos θ θ. () where a 1 is the length of link 1. All the link lengths are given in Table 1. Fig 1. System schematic. The interaction between system components. The desired path for the bucket tip is determined by an array of waypoints. Each waypoint consists of x, y and z coordinates of point O as illustrated in Figure. These waypoints can be converted into joint angles by using inverse kinematics equations presented by Koivo [ 5 ]. Equation 1 describes the angle of the excavator boom slew joint p oy θ = 1 tan, (1) pox where p oy is the y coordinate of point O and p ox the x coordinate. Equation gives the angle of the link around the Z 1 axis Link length (m) a 1 0.180 a 1.0 a 0.95 a 0.60 Table 1. Link lengths of the excavator boom. Equation gives the angle of the link around the Z axis. c sd θ = tan () s + cd a Finally, equation gives the angle of coordinate frame O around the Z axis θ ( π θ ) = θ + θ + θ b, (5) where θ b is the angle between the bottom of the bucket and the X axis. The digging angle θ describes the angle between the bottom of the bucket and horizontal ground plane. 5
Fig.. The assignment of the coordinate systems [ 5 ]. For the purposes of this small scale prototype, the bucket is kept perpendicular to the ground to mimic the orientation of the power mixer. Thus, determining the final position of the bucket tip requires three joint angles, namelyθ 1, θ, θ and the digging angle of the bucket θ. It directly follows that the x and y coordinates of the bucket ee are the same with joint. The bucket length thus defines the difference in the z coordinate. The waypoints are preloaded into the microcontroller memory via serial port. The reachability of each point is checked before coordinates are sent to ensure a predictable operation of the excavator. The reachable space of the excavator boom can be calculated when the link lengths and the minimum and maximum angles of each joint are known. The reachable space can be visualized by plotting a slice of the space according to Figure. This method is also very suitable for graphical user interface where the user can clearly see the reachable plane for a given z coordinate of the bucket tip. Fig. A slice of the reachable space of the excavator and an illustration of the orientation of all the boom links when the bucket tip is in coordinates (0.8, 1.5, 0.1). The area to be stabilized can be divided into rectangular pieces. The mixing head is kept at roughly constant depth as it is moved through the soil. The system parameters can be optimized if the largest rectangle for a given depth (z coordinate) is known. The dimensions of the rectangle with the largest area can be derived as and 1 x rect + = 8R k k 8R k (6) ( 8R + k k) 1 y rect =, (7) where R is the reach of the excavator and k is the radius of the unreachable area close to the excavator as illustrated in Figure. Fig. The rectangle inside a slice of the reachable space. The white area depicts area the excavator boom can reach for a given constant height of the bucket tip. The slew joint axis is at the origin. 5
. TEST PLAN Testing the accuracy and speed of the automatic operation is important for quantifying the possible performance gains in mass stabilization. To test the accuracy, a strategy for moving from one waypoint to the next must be devised. A waypoint strategy is characterized by the placement of the waypoints, e.g., the distance between the waypoints, and the tolerances associated with each waypoint. Each waypoint has a surrounding tolerance radius associated with it. If the resolvers indicate the bucket tip to be within the radius, the control algorithm then directs the bucket tip to move to the next waypoint. Modifying these factors will result in changes in performance measures. The goal was to find an optimal waypoint strategy within this control framework. To determine the value of the usable tolerances, the accuracy of the system has to be measured. Before the tolerances can be measured the resolvers must be calibrated. The system was calibrated by measuring the maximum and minimum readings for each resolver. The actual angles of the boom joints were then calculated by measuring the height of each joint from level floor and using trigonometric relationships to obtain the maximum and minimum joint angles corresponding to the resolver values.. RESULTS In order to determine the reliability of the resolver data, the real position of the bucket tip was measured from five different points and the results were compared with the position indicated by the resolver data and the kinematics equations. The comparisons are shown below in figures 5 and 6. The bucket joint angle θ was not included in the measurements. The bucket was always set so that the line from the bucket joint to the bucket tip was perpendicular to the ground. Fig 5. Box-Whisker plot for the joint angle error of the resolvers. The error in degrees of θ1 was between [0.5,.98]. θ had error between [0.65,.85] and the error of θ was between [-0., -6.]. Fig 6. Box-Whisker plot for position error of the bucket tip. The error in x-direction was between [-1.,.], for y-direction the error was between [-10,.8], and for z- direction [-1., 5.1]. The joint angles had errors less than 6 degrees for all the joint angles. The absolute error in distance was on the average 11. cm or 6.%. The standard deviation was.05 cm or.8%. The resolver data were concluded to be sufficiently accurate for this application based on the comparison. The calibration step is crucial for achieving high accuracy and these results could perhaps be further improved. 5. CONCLUSIONS The initial test results showed that the boom can be reliably moved within the reachable space of the excavator when no 5
external forces act on the boom. Thus the excavator is used in this research primarily as a test platform for automated boom movement. In mass stabilization the excavator joints have to tolerate large lateral loads. Currently, an important next step would be to develop a system for monitoring the forces applied to boom and the power mixer unit to avoid damage to the tool when encountering hard obstacles, such as rocks hidden in the soil. Beyond that, the problem of autonomous movement of the excavator and binding agent tank must also be solved to achieve full automation. 6. REFERENCES 1. Lahtinen, P. and Niutanen, V., evelopment of In-Situ Mass Stabilization Technique in Finland,. eep Mixing 009 Okinawa Symposium, International Symposium on eep Mixing & Admixture Stabilization Okinawa, Japan, May 19-1, 009.. Martikainen, J., Pahlsten J., Söderena P. & Ubiagege C., Automation in Mass Stabilization, 01, 8 th International AAAM Baltic Conference (Speech).. Kiviranta, J., Instrumentation of an Automated Excavator, Master s Thesis, Helsinki University of Technology, Espoo, 009.. Lee, C. S., Bae, J., and Hong,., Contour Control for Leveling Work with Robotic Excavator, Int. J. Precis. Eng. Man., 01, 1, 055 060. 5. Koivo, A. J., Kinematics of Excavators (backhoes) for Transferring Surface Material, J. Aerosp. Eng., 199, 7, 17. 7. CORRESPONING ARESS Panu Kiviluoma,.Sc. (Tech.) Aalto University School of Engineering, epartment of Engineering esign and Production P.O. Box 1100 00076 Aalto, Finland Phone: +58 50 8661 E-mail: panu.kiviluoma@aalto.fi http://edp.aalto.fi/en/ 8. AITIONAL ATA ABOUT AUTHORS Liukkonen, Jere, B.Sc. (Tech) Phone: +58 5 11 508 E-mail: jere.liukkonen@aalto.fi Knuuttila, Pekka, B.Sc. (Tech) Phone: +58 0 560 9767 E-mail: pekka.knuuttila@aalto.fi Nguyen, Tien, B.Sc. (Tech) Phone: +58 096 507 E-mail: tien.vannguyen@aalto.fi Ingale, Saurabh, B.Sc. (Tech) Phone: +58 1 8 068 E-mail: saurabh.ingale@aalto.fi Korkiala-Tanttu, Leena,.Sc. (Tech.), Professor Phone: +58 50 1 775 E-mail:leena.korkiala-tanttu@aalto.fi Kuosmanen, Petri,.Sc. (Tech.), Professor Phone: +58 500 8 81 E-mail:petri.kuosmanen@aalto.fi 55