A novel second-order online local reactive motion planner, "Right-of-Way-Based Higher-Order Velocity Obstacle Algorithm For Collision Avoidance In Multi-Agent Environments," is introduced in this thesis. It encompasses both deterministic and probabilistic frameworks and incorporates right-of-way rules mandated by regulatory authorities. Two distinct motion planning algorithms, namely "Right-of-Way-based Acceleration Velocity Obstacle" (R-AVO) and "Right-of-Way-based Probabilistic Acceleration Velocity Obstacle" (R-PAVO), are proposed.

The developed R-AVO algorithm considers kinematic constraints and reactive behaviour of other vehicles to generate collision-free, dynamically feasible, oscillation-free and goal-oriented trajectories for each unmanned vehicle (a.k.a. agent) in multi-agent, dynamic and obstacle-cluttered environments. Building upon the R-AVO method and the Bayesian Occupancy Filter for occupancy grid mapping, the R-PAVO algorithm is developed that considers the uncertainties, which are inherent in the perception systems, with the presence of uncertainties associated with information on other obstacles' and agents' motion and the occlusions caused by them. The inclusion of right-of-way rules fosters implicit coordination among agents with decentralization, mitigating the risk of reciprocal oscillation in their planned trajectories. Experiments and simulation studies are conducted to assess the functionality and effectiveness of the developed algorithms in diverse conflict scenarios. Extensive simulation studies are carried out to demonstrate performance in terms of relevant measures of effectiveness and real-time implementability of the developed algorithm.