Professor Sebastian Thrun will help us complete programming all the major systems of a robotic car over the course of seven weeks.
Join our team or become a mentor. We are meeting at Jigsaw Renaissance (Thursday evening and also during SCRoW: Soldering Coding and Robotics on Wednesday at 7pm).
At the last meeting, we had some highly experienced and enthusiastic people.
What we are given
- latitude and longitude of start point, destination, waypoints
- accurate to 4 decimal places
- might be places where GPS might not work (fall back to SLAM)
- Start -> Waypoint 1 -> Waypoint 2 -> Waypoint 3 -> destination
- Obstacle avoidance
- Cone detection (2 phase process)
- Surviving with weak lost GPS signals
- Be able to travel over a variety of diverse terrain such as grass, sand and concrete
- Optional/mandatory waypoints
Interesting points from the Rule book
- They use “datum WGS84” for GPS from Google Earth.
- Recording latitudes and longitudes that are accurate with respect to each other is more important than recording latitudes and longitudes that are accurate with respect to surveyed points. In other words, it’s more important a builder knows the distance and bearing between points than it is for them to accurately navigate to a surveyed location.
- Robothon Robomagellan 2006
- An example of the Robo-Magellan Robot, Sputnik (only robot to complete the course in 2007)
- Cone detection algorithm: Robomagellan Cone Searching/Tracking
High Level Algorithm
- Follow the path from start point to each way-point and finally to destination using GPS and SLAM
- Implement obstacle avoidance and control
- Image processing: detect triangle shaped orange color when the robot is within x distance from the waypoint
- Some way to check that the robot ‘touched’ the waypoint (touch sensors?)
A benefit of the having a grand goal is that we get exposed to a lot of technologies and products.
Here are some of the items that were mentioned:
- ESC Electronic Speed Control
- Differential GPS
- Basic4Android — use of android phones or tablets as robot brain
Other places or hackerspaces:
- Metrix Createspace
Real World Application:
Sebastian Thrun helped build Google’s amazing driverless car, powered by a very personal quest to save lives and reduce traffic accidents. Jawdropping video shows the DARPA Challenge-winning car motoring through busy city traffic with no one behind the wheel, and dramatic test drive footage from TED2011 demonstrates how fast the thing can really go.
In 2005, Stanley won $2,000,000 prize by winning the DARPA Grand Challenge, the largest prize money in robotic history. Stanley is an autonomous vehicle created by Stanford University’s Stanford Racing Team. The leader of Google and Stanford’s autonomous driving teams who created Stanley, Professor Sebastian Thrun is showing us how to program all the major systems of a robotic car at the Udacity, the free online course. Check out TED Talk – Sebastian Thrun: Google’s driverless car.