Autonomous Vehicles: Boundary Tracking and Control Laws

1
Autonomous Vehicles Boundary Tracking and
Control Laws

• By Jackie Brosamer
• June 19, 2008

2
Boundary Tracking

• Goal
• given a density field, we want the vehicle to
  find the boundary (where d(k) B) and track
  along it
• Basic algorithm
• rotate one way when inside the boundary and the
  opposite way when outside boundary

3
Problem Large initial crossing angle
4
Solution

• We modify our algorithm based on the time it
  takes to cross the boundary t and a predetermined
  reference angle

5
(No Transcript)
6
CUSUM filter

• But this modification relies on accurate record
  of the boundary crossing times, so we need to
  find a way to deal eliminate noise in our data
• We implement a cumulative sum algorithm to create
  a high side filter and a low side filter to
  increase accuracy

7
(No Transcript)
8
Multiple vehicle boundary estimation

• We assume the approximate boundary is a Markov
  model based on the state of the system s(k)

9
Optimization

• We then assume the boundary can be represented by
  an ellipse and formulate boundary estimation as
  an optimization problem

10
(No Transcript)
11
Steering Control Laws

• Steer by altering the curvature of the vehicles
  path

12
Coupling Control Laws

• Local coupling allows vehicles to follow other
  vehicles within their field of vision
• Also, stronger coupling between a designated
  leader and followers

13
Homotopy parameter

• Governs between global control law and local
  control law

14
Target Seeking

• Separate control law, does not guarantee swarming

15
Barrier Avoidance

• Average barrier direction given by
• We add the following term to steering control law