Lane-Keeping Control of Autonomous Vehicles Through a Soft-Constrained Iterative LQR
(2311.16900)Abstract
The accurate prediction of smooth steering inputs is crucial for autonomous vehicle applications because control actions with jitter might cause the vehicle system to become unstable. To address this problem in automobile lane-keeping control without the use of additional smoothing algorithms, we developed a soft-constrained iterative linear-quadratic regulator (soft-CILQR) algorithm by integrating CILQR algorithm and a model predictive control (MPC) constraint relaxation method. We incorporated slack variables into the state and control barrier functions of the soft-CILQR solver to soften the constraints in the optimization process so that stabilizing control inputs can be calculated in a relatively simple manner. Two types of automotive lane-keeping experiments were conducted with a linear system dynamics model to test the performance of the proposed soft-CILQR algorithm and to compare its performance with that of the CILQR algorithm: numerical simulations and experiments involving challenging vision-based maneuvers. In the numerical simulations, the soft-CILQR and CILQR solvers managed to drive the system toward the reference state asymptotically; however, the soft-CILQR solver obtained smooth steering input trajectories more easily than did the CILQR solver under conditions involving additive disturbances. In the experiments with visual inputs, the soft-CILQR controller outperformed the CILQR controller in terms of tracking accuracy and steering smoothness during the driving of an ego vehicle on TORCS.
Overview
-
The paper introduces the soft-constrained iterative linear–quadratic regulator (soft-CILQR) for smoother autonomous vehicle steering.
-
Soft-CILQR combines traditional CILQR with soft-constraint model predictive control to better handle state and input constraints.
-
Slack variables introduced in soft-CILQR allow for effective stabilization without needing additional smoothing techniques.
-
The algorithm outperformed traditional CILQR in simulations and vision-based experiments, showcasing smoother driving even with disturbances.
-
Soft-CILQR's fast computation makes it suitable for real-time applications in autonomous driving, improving safety and comfort.
Introduction to Soft-Constrained Iterative LQR for Autonomous Vehicles
Autonomous vehicles need to execute smooth steering actions for safe and stable driving, especially in challenging scenarios that involve lane-keeping on roads. Standard predictive control methods, which are commonly employed for this task, sometimes result in sharp, undesired inputs due to the presence of various constraints and disturbances, such as actuator limits and environmental noise.
Soft-Constrained Control Strategy
Recent advancements have led to the development of the soft-constrained iterative linear–quadratic regulator (soft-CILQR) algorithm. This method seeks to combine the conventional CILQR approach with a soft-constraint model predictive control (MPC) technique. By introducing slack variables into the optimization process, this new algorithm aims to soften constraints on the state and control inputs, thereby enabling the computation of stabilizing steering outputs more effectively. Importantly, this process involves modifying a vehicle’s lane-keeping control algorithm without the need for any after-the-fact smoothing algorithms or filters.
Comparison with Traditional Methods
The performance of the soft-CILQR algorithm was gauged against the traditional CILQR approach. The evaluations were divided into two parts: numerical simulations and experimental tests using a vision-based driving simulator. Numerical simulations showcased the soft-CILQR's ability to drive the system smoothly towards the reference state despite disturbances. During the vision-based manual experiments, both algorithms demonstrated satisfactory results in lane-keeping tasks on a simulated track. However, soft-CILQR displayed enhanced performance, with less conservative results and a smoother steering trajectory compared to the conventional CILQR, especially when noise was introduced into the system.
Findings and Implications for Self-Driving Technology
The findings indicate that the soft-CILQR can be an effective tool for maintaining the stability and comfort of autonomous vehicles even in the face of external disturbances. Its quick computational time suggests that it is suitable for real-time applications, a crucial aspect for in-the-moment decision-making required in autonomous driving.
In conclusion, the integration of slack variables within a CILQR framework allows for a more robust controller capable of handling the intricate dynamics of autonomous vehicle steering. This approach shows promise for enhancing self-driving technology, offering a path to more reliable and secure autonomous transportation.
Create an account to read this summary for free: