1: \begin{abstract}

2: {\it The problem of maneuvering a vehicle through a race course in minimum time requires computation of 

3:  both longitudinal (brake and throttle) and lateral (steering wheel) control inputs.    

4:  Unfortunately, solving the resulting nonlinear optimal control problem is typically computationally expensive and infeasible for real-time trajectory planning.

5:  This paper presents an iterative algorithm that divides the path generation

6:  task into two sequential subproblems that are significantly easier to solve. Given an initial path through the race track, the algorithm

7:  runs a forward-backward integration scheme to determine the minimum-time longitudinal speed profile, subject to

8:  tire friction constraints. With this fixed speed profile, the algorithm updates the vehicle's path by solving a convex optimization problem 

9:  that minimizes the resulting path curvature while staying within track boundaries and obeying affine, time-varying vehicle dynamics constraints.

10:  This two-step process is repeated iteratively until the

11:  predicted lap time no longer improves. While providing no guarantees of convergence or a globally optimal solution, 

12:  the approach performs very well when validated on the Thunderhill Raceway

13:  course in Willows, CA. The predicted lap time converges after four to five iterations, with each iteration over the full 4.5 km race course requiring

14:  only thirty seconds of computation time on a laptop computer. The resulting trajectory is experimentally driven at the race circuit with

15:  an autonomous Audi TTS test vehicle, and the resulting lap time and racing line is comparable to both a nonlinear gradient

16:  descent solution and a trajectory recorded from a professional racecar driver. The experimental results indicate that the proposed method is a viable option for 

17:  online trajectory planning in the near future.}

18: \end{abstract}

19: