Introduction to LiDAR

Transcription

1 Research Group Photogrammetry Department of Geodesy and Geoinformation (GEO) Vienna University of Technology Introduction to LiDAR Milutin Milenković with contributions from: Norbert Pfeifer, Camillo Ressel and Christian Briese

2 TU Wien Vienna University of Technology Mission technology for people 23,000 students, 5000 academics Faculties Mathematics and Geoinformation Physics Informatics Electrotechnical Engineering Civil Engineering Architecture and Spatial Planning Mechanical Engineering Chemistry Department of Geodesy and Geoinformation Photogrammetry and Remote Sensing Advanced Geodesy Geoinformation Cartography Engineering Geodesy Geophysics July Italy NEWFOR2014 Summer School 2

3 Research in LiDAR for modeling Measurement process Geometric calibration in airborne and terrestrial Lidar Radiometric calibration Models from lidar: DTM, buildings, wood volume, Lidar for biodiversity and forestry EU projects TransEcoNet ChangeHabitats2 Newfor Lidar for hydrology and hydraulics, geomorphology 1 st I.P.F. Lidar paper in 1997 Determination of Terrain Models in Wooded Areas with Airborne Laser Scanner Data [Kraus and Pfeifer, ISPRS-Journal, 1998] 1 st application of commercial FWF Lidar [Ullrich, Hollaus et al., SPIE Laser Radar Tech.&App., 2007] July Italy NEWFOR2014 Summer School 3

4 Overview July Italy NEWFOR2014 Summer School 4

5 Content Physical foundations (~30 slides) laser pulse laser equation radiometric calibration Geometric foundations (~15 slides) scanning principle direct georeferencing (GNSS, INS) Geometric quality control (~20 slides) point density precision accuracy July Italy NEWFOR2014 Summer School 5

6 Physical foundations July Italy NEWFOR2014 Summer School 6

7 Light Amplification by Stimulated Emission of Radiation LASER: coherent light, monochromatic, polarized, strongly bundeled (collimated) stimulated emission process by which, when perturbed by a photon, matter may lose energy resulting in the creation of another photon. The perturbing photon is not destroyed in the process, and the second photon is created with the same phase and frequency as the original. Frequency h Planck s constant same phase same wavelength Source: Wikipedia Optical (cavity) resonator: LASER medium choose one direction Mirror Mirror + Energy transparency July Italy NEWFOR2014 Summer School 7

8 Time = Distance Light (e.g. Laser) travels with the speed of light c group velocity, depends on composition of atmosphere: T, p, e c = m/s c ~ 3*10 8 m/s 1 ns (Nano second) = 10-9 s 1 ns = 30 cm d = t/2*c (Distance = time of flight (one way!) * speed of light) July Italy NEWFOR2014 Summer School 8

9 Pulsed Lasers time Laser pulses can be generated by various methods (gain-switching, Q- switching, mode-locking) Typical pulse duration in ALS 2 ns 10 ns 60 cm 3 m pulse length Source: Wagner et al, IAPRS XXX, July Italy NEWFOR2014 Summer School 9

10 Pulsed Lasers Pulse Repetition Rate (PRR) Pulse Repetition Frequency (PRF) number of emitted pulses per second Typical values in ALS: khz Example: PRR 100 khz Time between two pulses: 1/100k s = 10-5 s = 0.01 ms Distance between two pulses is 3*10 8 * 10-5 = 3000 m; while flying 1500m above the ground only one pulse is in the air. e.g. PRR = 100 khz, pulse width = 10ns (3m) 0.01 ms 10 ns time July Italy NEWFOR2014 Summer School 10

11 Wavelengths () in use dependent on laser medium and atmospheric windows: 8 μm 14 μm 3 μm 5 μm (mid IR) 0.7 μm 2.5 μm (near IR), e.g. 1.5 µm, 1064 nm 400 nm 700 nm (visible), e.g. 690 nm, 532 nm ultraviolet otherwise: absorption by water vapor, CO 2, O 3, July Italy NEWFOR2014 Summer School 11

12 Spectral signatures and laser wavelengths Reflectivity (in %) of water, dry soil, and vegetation Wavelength in use adapted to observed objects (e.g. vegetation and soil) No returns from water different wave length required for bathimetry (usually green lasers) July Italy NEWFOR2014 Summer School 12

13 values for 900nm, averaged over typical incidence angles Reflection coefficient for different materials Reflection coefficient depends in general on material reflectivity and angle of incident radiation and view point White paper up to 100% Dimension lumber (pine, clean, dry) 94% Snow 80-90% Beer foam 88% White masonry 85% Limestone, clay up to 75% Newspaper with print 69% Tissue paper, two ply 60% Deciduous trees typ. 60% Coniferous trees typ. 30% Carbonate sand (dry) 57% Carbonate sand (wet) 41% Beach sands, bare areas in dessert typ. 50% Rough wood pallet (clean) 25% Concrete, smooth 24% Asphalt with pebbles 17% Lava 8% Black neoprene 5% Black rubber tire wall 2% July Italy NEWFOR2014 Summer School 13

14 Laser beam width Beam widens due to diffraction and optical elements Beam waist (smallest diameter) w 0 Definition of beam width for beams with decreasing energy distribution from the center outwards (Gaussian beams) Energy drop to 1/e 2 Alternatively: Energy drop to 1/e, Beam radius expansion w(z) = w 0 (1+(z/(w 02 )) 2 ) ½ Divergence (full angle, beam width 1/e 2 ) is = 4//D D = 2w 0 Aperture [Jelalian 1992; Young 2000] typical values: < 1 mrad D 1 beam width 1/e 2 lateral energy distribution? w(z) z Energy Link: near field far field July Italy NEWFOR2014 Summer School 14

15 Energy distribution across (in area) Approximately Gaussian (Jutzi et al., 2003, wavelength 1543nm) Energy distribution Energy distribution along (in time) Approximately Gaussian Riegl -Puls, wavelength 1.5µm (Wagner et al., 2004) Footprint ( = 1 µm, D = 2 cm) of = 0.64 mrad 2000 m = 1.3 m Diffraction, /D Energy distributed in time and in area but concentrated in center July Italy NEWFOR2014 Summer School 15

16 Target Reflector (free) atmosphere Laser Range Finding (LRF) Pulse-Generator start Counter stop laser Transmitter t Detector Photodiode, Photomultiplier, Note on nomenclature Diffuse reflection = scattering Mirror reflection = specular reflection = reflection Transmission is not shown on this slide July Italy NEWFOR2014 Summer School 16

17 Laser Range Finding Range r = c * t / 2 with speed of light c ~ 3*10 8 m/s Necessary for successful measurement of t Diffuse reflection High SNR (signal to noise ration) Accuracy of time measurement s t leads to accuracy of range s r = 1.5*10 8 *s t, provided c is correct temperature change by 1 C 1 ppm = 1 mm/km Example: s t = 0.2 ns, s r = 3 cm Range is measured integral over the entire footprint July Italy NEWFOR2014 Summer School 17

18 P R PT 4R 2 4 A 2 4 Laser equation 1 4R ATM Received power P R [W] Transmitted power P T [W] Equally distributed along a sphere [m -2 ] Antenna gain (opening angle in respect to the sphere) [],db Area A of target [m 2 ] and with reflection coef. Equally distributed backscatter along a sphere [m -2 ] Backscatter in cone with opening angle Receiver aperture D [m 2 ] Atmospheric and system loss ATM, SYS [] Background radiation (sun, shot noise, ) P BK [W] 2 4 D 4 2 SYS P BK July Italy NEWFOR2014 Summer School 18

19 Laser equation P R PT 4R A 1 4R 2 4 D 4 2 ATM SYS P BK Received power P R should be high (SNR), how to achieve this? Transmitted power high Range small Target area large compared to beam divergence Target bright (no absorption, ) Receiver aperture big No background radiation (filters, avoid direct sun light) No atmospheric loss (e.g. Laser Altimetry at Mars, MOLA) July Italy NEWFOR2014 Summer School 19

20 Influence of target size on received power P R 2 PT D 2 4 R 4 4A ATM SYS P BK Influence of target size on received power Extended target A=R 2 T2 /4 P R 1/R 2 Example: open terrain Linear target A=R T d P R 1/R 3 Example: wire with diameter d (very small) Point target A=const P R 1/R 4 Example: leaf July Italy NEWFOR2014 Summer School 20

21 P R 2 PT D 2 4 R Backscatter cross section 4 4A σ ATM Backscatter cross section s [m²]: combines all relevant object parameters Isotrop =4 s = A Lambertian = s = 4A (orthogonal incidence) Mirror reflectors s = 0 General: SYS P BK A July Italy NEWFOR2014 Summer School 21

22 Multiple targets along the laser beam Within the beam: instantaneous field of view multiple different (area, linear, point) targets can be illuminated and can generate a sensable echo One pulse can therefore generate multiple echoes i.e.: first echo, intermediate echoes, last echo According to their arrival time, echoes are called first, second, intermediate, last echo Scanning over a forest, last echoes may be Reflections from the ground Reflections from dense vegetation above the ground July Italy NEWFOR2014 Summer School 22

23 Example: First and last echoes DSM of first echoes DSM of last echoes July Italy NEWFOR2014 Summer School 23

24 Discrete Systems and Range Resolution Objects spaced further than half the pulse length cause separated echoes Objects closer to each other cause compound (overlapping) echoes dt 1-2 dt 2-3 After: Briese et al. Digitale Geländemodelle im Stadtgebiet aus Laser- Scanner-Daten. Jahrestagung DGPF, Konstanz, July Italy NEWFOR2014 Summer School 24

25 Pulse interaction with objects = Echoes are widened Two echoes July Italy NEWFOR2014 Summer School 25

26 Discrete Echo vs. Full-Wave-Form LIDAR Echo waveform Discrete echoes Laser foot print: ~ 0.2 3m Recording the entire return signal using a sampling interval of ~1ns FWF ALS July Italy NEWFOR2014 Summer School 26

27 Discrete Echo vs. Full-Wave-Form LIDAR July Italy NEWFOR2014 Summer School 27

28 20 Full-Wave-Form 15 Formation Amplitude Emitted pulse Height distribution Distance (m) of objects Received echoes Amplitude Laser pulse Gaussian model Viewing direction Cross section [m 2 ] Amplitude Laser pulse Gaussian model Distance (m) Distance (m) Convolution Amplitude Distance (m) Data courtesy of Riegl Gmbh July Italy NEWFOR2014 Summer School 28

29 Full-Wave-Form-Analysis in Postprocessing Gaussian decomposition: Detection of echos by fitting Gaussian curves P wf ( t) c m i1 A e i ( t ) i 2 / s 2 i w The number of modes (echoes?) has to be estimated in advance (sets model complexity). The estimation is not linear, approximate values for unknowns The unknowns are: c, A,, s ; i 1, m i i i Cross section s R Information per echo: Amplitude (Intensity) P Range R Echo width w July Italy NEWFOR2014 Summer School 29

30 FWF Example Schönbrunn Example on next slides shows one ALS FWF strip over Schönbrunn castle (Vienna, Austria) acquired with a Riegl laser scanner. Orthophoto July Italy NEWFOR2014 Summer School 30

31 FWF Data Range Amplitude Echo width Ortho photo July Italy NEWFOR2014 Summer School 31

32 Radiometric Calibration: Motivation Amplitude [DN] Maria Theresia casern Vienna wide ALS campaign, December July Italy NEWFOR2014 Summer School 32

33 Radiometric Calibration: Incidence angle normalisation Laser equation in backscr. cross sect. s[m²]: Backscatter coefficient [db]: P R σ 4σ 2 Acos R 2 PT D 2 4 R 2 4 σ ATM SYS P BK Laser equation in : P R PT D 16R 2 2 ATM SYS P BK Beam cross section: Acos July Italy NEWFOR2014 Summer School 33

34 Ŝ s s Radiometric calibration Source: Wagner et al, IAPRS XXX, Reformulate Lidar equation: P r P t 16 D R 2 r 2 i atm sys i P t P r Ss ˆ i s Pˆ s p, i i D 2 r 16 Sˆ s s sys R 2 i Pˆ i atm s p, i P r... Received power [W] P t... Transmitted power [W] D r... Diameter of receiver aperture [m] R... Range [m] sys... system transmission factor atm... atmospheric transmission factor i... Backscattering coefficient [m 2 m -2 ] Ŝ... Amplitude of the system waveform [DN] s s... Standard deviation of the system waveform [s] P i... Amplitude of the i-th echo [DN] s p,i... Standard deviation of the i- th echo [s] C cal atm 10 2R a /10000 a... atmospheric attenuation coefficient [db/km] j July Italy NEWFOR2014 Summer School 34

35 Determination of C cal Calibration Target (CT) (e.g. Asphalt) CT, j 4 ~ CT ( ) cos j j CT,j... Backscattering coefficient of the CT of echo j [m 2 m -2 ] CT... reflectance of the calibration target CT j angle of incidence of the echo j within the calibration target Mean calibration constant C cal [m -2 s -1 ] for the ALS sensor C cal 1 N CT N CT 10 R 2R a / j 1 j j p, j j Pˆ s 4 ~ CT ( ) cos j j July Italy NEWFOR2014 Summer School 35

36 Amplitude (Vienna 2006) July Italy NEWFOR2014 Summer School 36

37 Diffuse reflectance (Vienna 2006) July Italy NEWFOR2014 Summer School 37

38 Geometric foundations July Italy NEWFOR2014 Summer School 38

39 Surface Sampling Scanning Different scanner mechanisms: Rotating one sided mirror (Fugro [1st system]) Rotating prisma mirror (Riegl) Rotating inclined mirror (Z+F-Laser, Faro, 3 rd -Tech) Pyramid mirror (Riegl) Oscillating mirror (Optech, Leica) Palmer scanner = nutating mirror (ScaLARS, NASA) Fiber scanner (TopoSys [1st systems]) July Italy NEWFOR2014 Summer School 39

40 Rotating mirror multiple facets Oscillating mirror Surface Sampling Scanning Fiber scanner Nutating mirror (Palmer Scanner) July Italy NEWFOR2014 Summer School 40

41 Flight direction Surface Sampling Scanning Rotating mirror Oscillating mirror Fiber scanner Nutating mirror or July Italy NEWFOR2014 Summer School 41

42 Scanning geometry Total opening angle typically Scan angle is (=/2) (up to 30 ) Swath width ψ sw 2h tan 2 Footprint diameter Fpinst 2 cos h inst July Italy NEWFOR2014 Summer School 42

43 Point density Influencial factors Flying height, h [m] Flying speed, v [ms -1 ] Effective(!) pulse frequency, PRF [Hz] Scan opening angle, [deg] Resulting average point density (avpd): avpd #points per second area per second PRF swathwidth v PRF 2h v tan ψ July Italy NEWFOR2014 Summer School 43

44 ALS data acquisition strip wise 12km Neighbouring strips have to overlap! Complete acquisition of the planned area Redundant observation (important for quality control) Optional means for increasing the point density (side overlap 50%) July Italy NEWFOR2014 Summer School 44

45 Georeferencing exterior orientation space module Airborne Laser Scanning is a dynamic (kinematic) data ground module acquisition method. Laser scanner observations are range and angle in its coordinate system. For each observed range-angle-pair per echo the position (x,y,z) in space, and the angular attitude (,,) of the sensor system have to be known. Exterior orientation (x,y,z,,,) determined by a Position and Orientation System (POS) GNSS (Global Navigation Satellite System) INS (Inertial Navigation System) aerial module July Italy NEWFOR2014 Summer School 45

46 ALS MSS Airborne laser scanning uses a multi sensor system (MSS) with these components: Laserscanner range measurement (LiDAR) angle measurement (scanner) GNSS reciever measures position of the plattform INS (Intertial Navigation System) measures rotation of the plattform synchronisation using the GNSS- Signal LIDAR GNSS INS July Italy NEWFOR2014 Summer School 46

47 GNSS Global Navigation Satellite System Purpose navigation (according to flight planning) positioning of the sensor synchronization of measurements! Instrument antenna on top of aircraft receiver inside ground module Method kinematic differential GNSS Measurment frequency: 1-10Hz Absolute Precision: Planar: ±5-10cm Height: ±8-15cm IGI, AEROcontrol July Italy NEWFOR2014 Summer School 47

48 Inertial Navigation System Purpose attitude determination of the sensor platform interpolation of position between GNSS measurements Instruments gyroscopes accelerometers Measuring angular and acceleration increments Frequency: Hz Precision: ± m Applanix POS-AV IMU July Italy NEWFOR2014 Summer School 48

49 ALS-System Oscilating Mirror Lidar-Sensor (Head) incl. INS Glass Fiber Cable GPS/IMU Control-Unit Digital Camera GPS Antenna Power Supply and Storage Laser Generator and Processor July Italy NEWFOR2014 Summer School 49

50 Sensors July Italy NEWFOR2014 Summer School 50

51 6-parameter trajectory as function of time Aim: Computation of the platform exterior orientation (position + angles, x,y,z,,,) for each laser shot (pulse emmission) GPS operated in post processing with reference stations ( ) Drift of INS limits length of strips INS + GPS data integrated in Kalman filter Precision of the trajectory: Position 5cm 10cm (Elevation x 1,5), Angles: ~0.01 t 1 z p 0 t 2 z x y x t 3 z x p 0 y July Italy NEWFOR2014 Summer School 51 p 0 y

52 Mathematical model for ALS (direct georeferencing) GNSS IMU, Surface point (in the reference system) GNSS antenna: phase centre Rotation from sensor body frame to reference system Range and deflection angle of the laser beam Mounting offset vector ( lever arm ) Mounting rotation bias ( misalignment ) LiDAR Fig. adapted from Cramer 2001 Involved sensors (GNSS, IMU, LiDAR) have systematic errors (constant over a certain period of time; e.g. minutes or days), which generate errors in X. Especially the mounting calibration is often not known accurately enough July Italy NEWFOR2014 Summer School 52

53 Point cloud! (x i,y i,z i ) in R 3 + Intensity + Echo Info + FWF Info

54 Airborne Laser Scanning (ALS) Sensors Technical data of typical ALS Systems: Specification Wavelength Pulse Duration Beam Divergence Pulse Repetition Rate Field of View Scan Rate Scan Pattern Footprint Multiple Returns Intensity Operating Altitude GPS Frequency INS Frequency Accuracy (elevation) Accuracy (planimetric) Typical Value 1.0 and 1.5m 5-15ns 0.2-2mrad 1-266kHz Hz Zigzag, parallel, elliptical, sinusoidal m 2-8 or full-waveform Yes m 1-2Hz (max. 10 Hz) Hz 3-10cm cm July Italy NEWFOR2014 Summer School 54

55 Quality Control July Italy NEWFOR2014 Summer School 55

56 What accuracy is possible? Accurac check using control DTMs derived from 816 check points: Flight Vienna 1999, 500m AGL, 0.5x1m point density 2 1 Region RMS [cm] б [cm] All points ± 10.5 ± 7.1 Park, dense vegetation ± 14.5 ± 11.1 Park, low vegetation ± 11.4 ± 7.8 Park, open area ± 8.6 ± 4.5 Street with parking cars ± 9.2 ± 3.7 Street without cars ± 2.4 ± City of Vienna, MA41, Vienna July Italy NEWFOR2014 Summer School 56

57 Project Parameters: Flight Date (I) airplane borne, 4 pts. / m², September Digital Surface Model (DSM) Digital Terrain Model (DTM) July Italy NEWFOR2014 Summer School 57

58 Project Parameters: Flight Date (II) helicopter borne, 8 pts. / m², March Digital Surface Model (DSM) Digital Terrain Model (DTM) July Italy NEWFOR2014 Summer School 58

59 Project Parameters: Point Density 271 control points: point distance: ~1m s = ±18cm point distance : ~3.1m s = ±29cm Point density is the most important parameter regarding accuracy and level of detail (e.g. building reconstruction) July Italy NEWFOR2014 Summer School 59

60 What are the main quality parameters of ALS data from a geometric point of view? The geometric quality of ALS data and of the models derived from them (e.g. DxM, buildings, trees, ) depends on many parameters: Point density It is the most important parameter for the level of detail (and thus for the level of accuracy) of the derived models (e.g. terrain- or building models) Precision (random errors, noise) Delimited by noise of the system; e.g. standard deviation of points of 1 strip with respect to a plane (depends also on the strip s interior orientation (laser sensor, IMU, GNSS)) Accuracy Relative: consistency of the points from several overlapping strips (how well do 2 strips fit together?) (defined by relative orientation of the block of strips) Absolute: consistency of the points from the entire block of strips with respect to control data (how well does the block of strips fit in to the world system?) (defined by absolute orientation of the block) Are the point density, precision and accuracy of the delivered ALS data as planned /expected? ALS quality control July Italy NEWFOR2014 Summer School 60

61 Point density July Italy NEWFOR2014 Summer School 61

62 Point density Example: Point density: s Average number of points per square meter (in case of multiple echoes only the last echo should be counted) within an analysis unit. Point density in a single strip needs to be distinguished from the density in the entire block of strips. Areas with no data need to be dealt with separately. density = 6/s 2 The analysis units are squares with side s; e.g. s = 5m. Side s should be chosen so that enough points are inside the analysis unit July Italy NEWFOR2014 Summer School 62

63 Example ALS-Vaihingen: point density of single strip Point density (Str. 5) [pnts/m 2 ], last echos, analyse unit: 5x5m m 1.2 m July Italy NEWFOR2014 Summer School 63

64 Example ALS-Vaihingen: point density of block Point density (Str. 5) [pnts/m 2 ], last echos, analyse unit: 5x5m 2 Color coding using 4 pnts/m 2 as reference July Italy NEWFOR2014 Summer School 64

65 Example ALS-Vaihingen: point density of block Point density (Str. 5) [pnts/m 2 ], last echos, analyse unit: 5x5m 2 Color coding using 3 pnts/m 2 as reference July Italy NEWFOR2014 Summer School 65

66 Precision July Italy NEWFOR2014 Summer School 66

67 Precision Precision (random errors, noise) Delimited by noise of the system; e.g. standard deviation of points of one strip with respect to a plane (depends also on the strip s interior orientation (laser sensor, IMU, GNSS)) Method Interpolate digital surface model (DSM) from the given ALS points using moving planes interpolation and check the standard deviation s 0 of that interpolation in smooth areas. The excentricity is the distance between the CoG of the ALS points used for the interpolation and the raster point. will be used later. Moving planes interpolation s 0 Difference: s Z vs. s 0 s Z July Italy NEWFOR2014 Summer School 67

68 Precision: Example Sigma0 of moving planes interpolation opalszcolor -scale 0,0.5 -ncl 20 #Used: Mean: Std: Median: SigMad: rough regions clearly visible roughness mask smooth := s 0 < 0.15 smooth := s 0 < 0.15 #Used: Mean: Std: Median: SigMad: July Italy NEWFOR2014 Summer School 68

69 Accuracy July Italy NEWFOR2014 Summer School 69

70 Accuracy Relative Measure for the consistency of the points from several overlapping strips. How well do the strips fit together? (defined by relative orientation of the block of strips) Analyze the difference of overlapping strips. Benefit: relative accuracy can be checked within the data alone (provided ALS strips have enough overlap) Absolute Measure for the consistency of the points from the entire block of strips with respect to control data. How well does the block of strips fit in to the world system? (defined by absolute orientation of the block) Drawback: Checking the absolute accuracy requires some sort of ground truth. The latter, however, is not simple to get hand of (expensive and elaborate) July Italy NEWFOR2014 Summer School 70

71 Relative accuracy via strip differences Relative orientation of the strips. How well do the strips fit together? Checked by : difference of overlapping ALS strips July Italy NEWFOR2014 Summer School 71

72 Strip difference: Example [m] NEWFOR2014 Summer School 72

73 Strip difference: with/without mask Unmasked strip difference: with vegetation Masked strip difference: without vegetation July Italy NEWFOR2014 Summer School 73

74 Strip difference: Appearance of buildings Color-coded strip difference: [m] effect of strip differences at buildings: strip 1 strip 2 mainly planar errors (caused by wrong georeferencing; mainly due to errors of the mounting calibration) 0.5m Schönbrunn 2004 Str 8/ July Italy NEWFOR2014 Summer School 74

75 Improving the Georeferencing of the ALS strips So far: Color-coded strip differences: Benefit: continuous representation of the relative geometric accuracy using height differences ( Look for suspicious patterns, e.g. buildings or roll effects ) Drawback: no quantitative information about the (relative) planar accuracy (and before strip adjustment also not about height, because differences are not computed at corresponding locations!) Solution: LSM at manually selected locations ( discrete XYZ displacements between two strips) (LSM = Least Squares Matching); July Italy NEWFOR2014 Summer School 75

76 Improving the Georeferencing of the ALS strips Original: Improved: In cooperation with Vermessung Wenger-Öhn & AVT July Italy NEWFOR2014 Summer School 76

77 Thank you! July Italy NEWFOR2014 Summer School 77