MPC-BicycleModel

This is Team 2's final project git repository for ROB535: Self-Driving Cars.

In this project, we control a 6-state bicycle to race while avoiding random obstacles using MPC.

The team members include: Ziqi Han, Siyuan Yin, Qilin He, Yifan Wang, Yuzhou Chen.

Project details and task requirements can be found here.

---

1. Usage of our repo

Add directories src and trajectories to path.

execute run.m

To generate gif for visualization:

execute make_gif.m

2. Result

• For task1:

Rank	Percentage of Completion(%)	Time consumption(sec)	Group number
1	100	4.8777	7
2	100	4.9119	9
3	100	5.185	12
4	100	7.266	6
5	100	10.7061	3
6	100	16.4615	1
7	100	28.9013	11
8	100	34.1071	10
9	100	49.0447	8
10*	100	50.3462	Our Team
11	100	68.3962	4
12	100	260.4474	5

• For task2:

Rank	Percentage of Completion(%)	Average t_score(points)	Group number
1	100	168.706	12
2*	100	278.608	Our Team
3	100	280.08	7
4	100	297.528	6
5	100	311.648	9
6	100	341.856	10
7	100	436.214	4
8	100	598.408	3
9	100	734.988	8
10	76.356	-	5
11	5.97218	-	11
12	0.11138	-	1

3. Documentation

• Report of Ziqi Han
• Report of Siyuan Yin
• Report of Qilin He
• Report of Yifan Wang
• Report of Yuzhou Chen

4. Project Log & Todo List

• 1. Finish writing PID control for part1.

• 2. Use PID control for part2.

  • 2.1 Debug.
  • 2.2 Test for success rate and time => This method is abandoned.

• 3. Use fmincon for part2.

  • 3.1 Finish writing constraints.
  • 3.2 Debug => This method is abandoned.

• 4. Use MPC for part2.

  • 4.1 Generate reference trajectory for MPC.
  • 4.2 Debug.
  • 4.3 solve QP no solution problem.
  • 4.4 Test for success rate and time.

• 5. Upload and Report.

  • 5.1 Draw trajectory.
  • 5.2 Write Report and Upload code.

• 6. Use MPPI for part2.

  • 6.1 Write Cost function.
  • 6.2 Write MPPI loop.
  • 6.3 Test for success rate and time.