Development of a Novel Steering Control Collision Avoidance System Open Access
Downloadable ContentDownload PDF
In this research development of a steering collision avoidance system was investigated. First, different control approaches to perform an evasive collision avoidance maneuver using active steering was presented. Both linear and nonlinear controllers to control the combined lateral and longitudinal motion of the vehicle using predefined trajectories are compared. A proportional-derivative controller, a linear quadratic regulator (LQR), and two different sliding mode controllers (SMC) were developed. The second SMC model includes an additional velocity error term, which augments the model with a steering actuator term. The controllers were implemented on a bicycle model and a 17 DOF vehicle model. The results showed that all controllers perform similarly in controlling the trajectory of the bicycle model. However, in implementation on the non-linear full vehicle dynamics model, the LQR and SMCs provided similar position tracking but the two SMCs performed better in minimizing the Yaw (directional) error at the end of the trajectory. In addition, the enhanced SMC provided a smoother steering angle profile, which is a desirable feature.Moreover, the possibility of performing severe collision avoidance maneuvers using trajectory optimization is investigated. A two degree of freedom vehicle model was used to represent dynamics of the vehicle. First, a linear tire model was used to calculate the required steering angle to perform the desired evasive maneuver, and a neighboring optimal controller was designed. Second, direct trajectory optimization algorithm using collocation technique was used to find the optimal trajectory with a nonlinear tire model. To evaluate the results, the calculated steering angles were fed to a full vehicle dynamics model. It was shown that the neighboring optimal controller was able to accommodate the introduced disturbances. Comparison of the resultant trajectories with other desired trajectories showed that it results in a lower lateral acceleration profile and a smaller maximum lateral acceleration, thus the time to perform an obstacle avoidance maneuver can be reduced using this method. A simulation case study of a limited lateral acceleration with constrained direct trajectory optimization showed that using the proposed trajectory optimization technique requires less time than that of trapezoidal acceleration profile for a lane change maneuver.Furthermore, a steering control collision avoidance system was designed and implemented in a car driving simulator. This system was tested on 20 human subjects. Analyzing the surveys from the subjects and the experimental data showed that the system is highly effective in reducing number of accidents of the introduced near collision encounters; and the subjects liked the system and the way that it was activated.