![Page 1: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/1.jpg)
Apprenticeship Learning for Motion Planning with Application to Parking Lot
Navigation
Presented by: Paul Savickas
![Page 2: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/2.jpg)
Outline
Motivation Background The Problem Methods and Model Results Discussion
![Page 3: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/3.jpg)
Motivation
In 2005: ~6,420,000 Auto Accidents in the US Total financial cost >$230 Billion Injuries ~2.9 Million 42,636 Deaths
– 115 per day– 1 every 13 minutes
Source: http://www.car-accidents.com/pages/stats.html
![Page 4: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/4.jpg)
Background
DARPA Urban Challenge Stanford Racing Team Junior
![Page 5: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/5.jpg)
Stanford’s “Junior”
http://www.youtube.com/watch?v=BSS0MZvoltw&NR=1
![Page 6: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/6.jpg)
The Problem
Motion/Path planning algorithms are complex– Many parameters– Hand tuning required
How can we simplify this process?
![Page 7: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/7.jpg)
Methods
Path-Planning and Optimization– Sequence of states– Potential-field terms– Weights
![Page 8: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/8.jpg)
Apprenticeship Learning
Learn parameters from expert demonstration Useful when no reward function available
Example: Teaching a person to drive
![Page 9: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/9.jpg)
Model
Parameters– Length– Length (Reverse)– Direction Changes– Proximity to Obstacles– Smoothness– Parking Lot Conventions– Lane Conventions
![Page 10: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/10.jpg)
Parameter Tuning
Switching Directions Penalize offroad
Distance from lane
Penalize Switching
Follow Directions
![Page 11: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/11.jpg)
Experiment
Human Driver demonstration– “Nice”– “Sloppy”– “Reverse-allowed”
Autonomous Navigation using learned parameters
![Page 12: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/12.jpg)
Results – “Nice”
![Page 13: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/13.jpg)
Results – “Sloppy”
![Page 14: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/14.jpg)
Results – “Reverse-allowed”
![Page 15: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/15.jpg)
Results – Interesting Anomalies
![Page 16: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/16.jpg)
Discussion
How do these results apply to other problems?
Determination of parameters is still an issue.
Why would we want to drive “sloppy”?..
Where do we go from here?
![Page 17: Apprenticeship Learning for Motion Planning with Application to Parking Lot Navigation Presented by: Paul Savickas](https://reader035.vdocuments.us/reader035/viewer/2022081515/56649d765503460f94a5817f/html5/thumbnails/17.jpg)
Video
DARPA Urban Challenge http://www.youtube.com/watch?v=P0NTV2m
bJhA&feature=related
Junior’s Results http://www.youtube.com/watch?v=xcNFUi06f
h8