Autonomous navigation requires real-time trajectory optimization with limited onboard computational resources, where traditional optimization-based methods often impose heavy computational burdens and require extensive parameter tuning. To achieve transparent navigation results and minimize computational burden, this paper introduces an iterative learning planning (ILP) approach for navigation. By dynamically integrating the control layer into the planning layer, ILP can achieve efficient trajectory optimization through functional iterative learning, providing transparent results that enhance navigation efficiency. This approach significantly improves real-time performance, achieving a time complexity ofO(k*n), wherek*represents the number of iterations andnrepresents the number of waypoints. Simulations and quadcopter experiments are conducted to verify the proposed framework. Results show that ILP significantly improves computational efficiency, ensures safe navigation, and maintains robustness against disturbances, demonstrating its potential as a practical solution for real-world autonomous systems.

This paper presents a novel model-free spatial iterative learning (IL) framework to enhance the efficiency of vector field (VF) navigation for mobile robots. By integrating the idea of iterative learning control (ILC) with VF, this framework utilizes historical data to enhance navigation efficiency significantly, reducing traversal time and expanding the applicability of IL to rapid navigation. Importantly, it has low time complexity with$O(n)$per iteration, where$n$denotes the waypoints number, preventing the significant computational overhead caused by the increasing waypoints in existing methods, which often exceeds$O(n^{2})$, making it well-suited for real-time planning. Moreover, the approach is inherently model-free, leaning on historical data, thus enabling agile navigation with limited reliance on intricate model details. The paper presents a comprehensive theoretical analysis of the stability, time optimality, time complexity, parameter insensitivity, robustness, and usage. Extensive simulations and experiments highlight its efficiency, promising a transformative impact on mobile robot navigation through the proposed IL.

It is often necessary for drones to complete delivery, photography, and rescue in the shortest time to increase efficiency. Many autonomous drone races provide platforms to pursue algorithms to finish races as quickly as possible for the above purpose. Unfortunately, existing methods often fail to keep training and racing time short in drone racing competitions. This motivates us to develop a high-efficient learning method by imitating the training experience of top racing drivers. Unlike traditional iterative learning control methods for accurate tracking, the proposed approach iteratively learns a trajectory online to finish the race as quickly as possible. Simulations and experiments using different models show that the proposed approach is model-free and is able to achieve the optimal result with low computation requirements. Furthermore, this approach surpasses some state-of-the-art methods in racing time on a benchmark drone racing platform. An experiment on a real quadcopter is also performed to demonstrate its effectiveness.