Collision-free Navigation of Mobile Ground Robots and UAVs in Complex Environments with Static and Dynamic Obstacles

This thesis presents novel navigation strategies for mobile robots and unmanned aerial vehicles (UAVs) operating in complex, dynamic environments, with a focus on indoor applications and sensorless robot architectures. Traditional autonomous navigation relies heavily on onboard sensing and computational resources, which can be costly, power-intensive, and limited in perception range. In contrast, this work explores hybrid and infrastructure-assisted navigation methods that decouple sensing, trajectory planning, and low-level control. The first part of the research develops hybrid navigation frameworks for nonholonomic UAVs and mobile robots, integrating global path planning in known environments with reactive control algorithms for real-time obstacle avoidance. These methods enable safe traversal through environments containing both static and dynamic obstacles for low computational robots, validated through high-fidelity simulations. The second part introduces an innovative sensorless navigation paradigm in which fixed sensor networks and off-board computing facilities replace onboard sensors. The fixed infrastructure continuously localises robots, tracks moving obstacles, and generates collision-free trajectories. Advanced multi-object tracking techniques, including the Interacting Multiple Model Joint Probabilistic Data Association (IMM-JPDA) filter, ensure robust tracking in cluttered environments. This approach supports both single-robot and multi-robot scenarios, enabling coordinated navigation in shared spaces. The final contributions include real-time trajectory replanning algorithms for sensorless robots, leveraging fixed LiDAR networks to optimise paths for user-defined metrics such as minimal distance and reduced privacy intrusion. Across all proposed systems, comprehensive MATLAB-based simulations and hardware implementation demonstrate the effectiveness of the methods in ensuring safe, efficient, and scalable navigation in dynamic indoor environments. The results establish that infrastructure-assisted navigation offers a cost-effective, computationally efficient, and scalable alternative to fully onboard autonomy, paving the way for practical deployment in applications such as warehouse automation, hotel service robots, and collaborative multi-robot systems.