Path planning algorithm based on pulse-coupled neural network with dual constraints of heading-deviation and node penalty

Path planning for mobile robots is a crucial aspect of robot navigation. Traditional search algorithms are limited by heuristic design and search strategies, often struggling to balance search efficiency and path quality in intricate obstacle environments. The Pulse-Coupled Neural Network (PCNN) offers inherent parallelism and completeness due to its autowave propagation mechanism. Nevertheless, the lack of directional guidance in the neuron activation process necessitates the activation of a significant number of free grid cells on extensive maps, thereby impeding search efficiency. This paper introduces a Dual-Constrained PCNN path planning algorithm to tackle the aforementioned issues. The algorithm incorporates two constraint mechanisms within the PCNN autowave propagation framework. The first mechanism involves a penalty term for heading deviation, which adjusts the internal activity of neurons based on the angle deviation between the current and target directions. This adjustment facilitates the autowave propagation toward the desired direction. The second mechanism involves a node penalty term that adjusts the dynamic threshold of neurons, which guides the path to converge precisely toward the goal. By synergistically integrating these two constraints, the algorithm effectively reduces redundant activation, improves the PCNN autowave propagation process, and ultimately enhances planning efficiency. To assess the proposed method’s generalization ability across diverse environments, a comprehensive experimental setup was devised, encompassing varied map dimensions, obstacle densities, and obstacle configurations. Results indicate that in environments with sparse obstacles, DC-PCNN demonstrates consistent search success rates across maps of different sizes, with path-planning runtime increasing linearly with map dimensions. The resulting path closely aligns with the straight-line distance between the start and goal points, featuring minimal turns and smooth trajectory, underscoring its robust cross-scale generalization and search efficiency. Conversely, in high-density obstacle settings, the fixed-parameter DC-PCNN exhibits a reduced success rate attributed to excessive directional bias, leading to local entrapment and goal failure. Upon implementing a parameter adaptive strategy, the method’s success rate in high-density scenarios significantly improves, affirming the efficacy of the adaptive mechanism.