TABLE 3.

Characteristics of reinforcement learning based path planning and collision avoidance.

Characteristic	Description
Objective	Find safe, efficient paths and avoid collisions
State Representation	Environment states, possibly occupancy grid
Action Space	Movement actions or possible trajectories, sensing actions
Reward Function	Penalty for collisions, rewards for avoidance and goal reaching
Observation	Sensor data for obstacle detection (e.g., vision or LiDAR)
Challenges	High-dimensional states, real-time constraints and dynamic obstacles
Citations	Larsen et al. (2021); Popovic et al. (2020); Rückin et al. (2022); Chen et al. (2017); Yanes Luis et al. (2022); Choi et al. (2021); Woo and Kim. (2020)