An adaptive ant colony optimisation for improved lane detection in intelligent automobile vehicles

This paper presents an improved lane detection algorithm using an adaptive ant colony optimisation (aACO) based edge detection technique. In the paper, we first modified the ACO to select its control parameters (pheromone influencer, heuristic influencer, evaporation rate and the pheromone decay coefficient) adaptively. The modified ACO was first used to solve some selected CEC benchmark functions of diverse properties. Thereafter, the algorithm was used to implement an edge detection algorithm for lane detection. To this, a threshold tolerance technique was developed to determine the element of the final ant pheromone matrix that constitutes a lane edge. Three lane detection testbeds (traffic, liquid and cloudy) were created to evaluate the performance of the develop algorithm. Simulations were performed using MATLAB and results shows that the dynamic parameter-based lane detection can detect lane line better than the Canny edge detector, irrespective of any occlusion in the lane images.

Automated summary

This paper presents an improved lane detection algorithm using an adaptive ant colony optimisation (aACO) based edge detection technique. The algorithm can detect lane lines better than the Canny edge detector, irrespective of any occlusion in the lane images.