This article studies how to improve the generalization performance and learning speed of the navigation agents trained with deep reinforcement learning (DRL). Although DRL exhibits huge potential in robot mapless navigation, DRL agents performing well in training scenarios are often found to perform poorly in unfamiliar scenarios. In this work, we propose that the representation of LiDAR readings is a key factor behind the degradation of agents' performance and present a powerful input preprocessing (IP) approach to address this issue. As this approach uses adaptively parametric reciprocal functions to preprocess LiDAR readings, we refer to this approach as IPAPRec and its normalized version as IPAPRecN. IPAPRec/IPAPRecN can highlight important short-distance values and compress the range of less-important long-distance values in laser scans, which well address the issues induced by conventional representations of laser scans. Their high performance was validated by extensive simulation and real-world experiments. The results show that our methods can substantially improve navigation agents' generalization performance and greatly reduce the training time compared to conventional methods.

Deep reinforcement learning (DRL) demonstrates great potential in mapless navigation domain. However, such a navigation model is normally restricted to a fixed configuration of the range sensor because its input format is fixed. In this letter, we propose a DRL model that can address range data obtained from different range sensors with different installation positions. Our model first extracts the goal-directed features from each obstacle point. Subsequently, it chooses global obstacle features from all point-feature candidates and uses these features for the final decision. As only a few points are used to support the final decision, we refer to these points as support points and our approach as support point-based navigation (SPN). Our model can handle data from different LiDAR setups and demonstrates good performance in simulation and real-world experiments. Moreover, it shows great potential in crowded scenarios with small obstacles when using a high-resolution LiDAR.

In the domain of maneuvering target tracking, conventional algorithms frequently utilize multiple predefined mathematical models to approximate target motion. Nonetheless, the intrinsic randomness and unpredictability of maneuvering targets present significant challenges for accurate motion modeling. To address these challenges, this paper proposes a novel Transformer-Aided Multiple Model (TAMM) algorithm for maneuvering target tracking. The primary innovation of TAMM algorithm is the integration of a Transformer-Aided Recognition Module (TARM), which significantly enhances adaptability to the frequent state changes encountered in target tracking. Additionally, a dynamic adjustment strategy is implemented to optimize the tracking process, aligning the most suitable motion model and turn rate with each phase, guided by the TARM’s output. Extensive simulation results affirm that the proposed algorithm outperforms the traditional UKF algorithm and the Interactive Multiple Model (IMM) algorithm, as well as the DeepMTT algorithm and the TBN algorithm based on deep learning in terms of stability and localization accuracy across diverse scenarios characterized by varying positions, velocities, and turn rates.

In this letter, we present a deep reinforcement learning-based dimension-configurable local planner (DRL-DCLP) for solving robot navigation problems. DRL-DCLP is the first neural-network local planner capable of handling rectangular differential-drive robots with varying dimension configurations without requiring post-fine-tuning. While DRL has shown excellent performance in enabling robots to navigate complex environments, it faces a significant limitation compared to conventional local planners: dimension-specificity. This constraint implies that a trained controller for a specific configuration cannot be generalized to robots with different physical dimensions, velocity ranges, or acceleration limits. To overcome this limitation, we introduce a dimension-configurable input representation and a novel learning curriculum for training the navigation agent. Extensive experiments demonstrate that DRL-DCLP facilitates successful navigation for robots with diverse dimensional configurations, achieving superior performance across various navigation tasks.

Deep reinforcement learning (DRL) has exhibited considerable promise in the training of control agents for mapless robot navigation. However, DRL-trained agents are limited to the specific robot dimensions used during training, hindering their applicability when the robot's dimension changes for task-specific requirements. To overcome this limitation, we propose a dimension-variable robot navigation method based on DRL. Our approach involves training a meta agent in simulation and subsequently transferring the meta skill to a dimension-varied robot using a technique called dimension-variable skill transfer. During the training phase, the meta agent for the meta robot learns self-navigation skills with DRL. In the skill-transfer phase, observations from the dimension-varied robot are scaled and transferred to the meta agent, and the resulting control policy is scaled back to the dimension-varied robot. Through extensive simulated and real-world experiments, we demonstrated that the dimension-varied robots could successfully navigate in unknown and dynamic environments without any retraining. The results show that our work substantially expands the applicability of DRL-based navigation methods, enabling them to be used on robots with different dimensions without the limitation of a fixed dimension.