Articulated objects are ubiquitous in daily environments, and effective manipulation of these objects is essential for advancing open-world robotics. Existing approaches, which rely heavily on large-scale data collection or simulation, often face limitations in real-world applications, including issues with generalization and the sim-to-real gap. In this paper, we introduce the Automated Articulated Object Parameter Learning (AAOPL) framework, which autonomously learns the articulation parameters of real-world articulated objects through direct interaction. This approach enables robots to generate precise manipulation trajectories without relying on predefined object models or extensive human demonstration data. To accelerate the learning process, we develop Accelerated Single-Step Gradient (ASSG) algorithm, which efficiently refines the articulation parameters by leveraging real-time execution feedback. Experimental results demonstrate that AAOPL can learn accurate articulation parameters within 30 minutes (80 trials) and generate robust manipulation trajectories, outperforming baseline methods in terms of both task completion and force efficiency. Our approach eliminates the need for large-scale training datasets and can adapt to various articulated objects in real-world environments, offering a scalable solution for autonomous robotic manipulation in unstructured settings.

The data-driven paradigm has shown great potential in solving many decision-making tasks. In the robot navigation realm, it also sparked a new trend. People believe powerful data-driven methods can learn efficient and general navigation policies from a vast offline dataset. However, robot navigation tasks differ from common planning tasks and present unique challenges. It often involves multi-objective optimization to meet arbitrary and ever-changing human preferences. It should also overcome the short-sighted problem to obtain globally optimal performance. Furthermore, high planning frequency is needed to address real-time demands. These factors obstruct the application of data-driven methods in robot navigation. To address these challenges, we integrate one of the most powerful data-driven methods, the diffusion model, into robot navigation. Our proposed approach, NaviDiffuser, utilizes a novel classification label to guide the diffusion model in capturing the complex connections between navigation and human preferences. Its Transformer network backbone outputs action sequences to alleviate short-sightedness. It also includes special distillation skills to boost the planning speed and quality. We conduct experiments in both simulated and real-world scenarios to evaluate our approach. In these experiments, NaviDiffuser not only demonstrates an extremely high arrival rate but also adjusts its navigation policy to align with different human preferences.

Dual-arm robotic systems hold great potential for complex bimanual tasks that require intricate and coordinated manipulation, such as holding and transporting a tray with a cup of coffee while navigating through cluttered environments. However, these tasks pose significant challenges due to the inherent closed-chain constraints between the arms and the object, as well as the need for real-time collision avoidance, especially in real-world applications. To address these challenges, we introduce a hierarchical framework that combines learning-based planning with classical control theory to ensure whole-body collision avoidance movement while maintaining the kinematic relationship. In addition, we present a novel, efficient, and cost-free data generation method specifically designed for dual-arm cooperative tasks, overcoming the lack of sufficient training data. Extensive experiments in both simulation and real-world scenarios demonstrate that our approach improves the success rate by 26.3% compared to existing planning methods and by 54.7% compared to end-to-end methods. These results highlight the advantages of our method in whole-body collision avoidance and environmental adaptability, making it a promising solution for dual-arm cooperative tasks.

Reinforcement learning has shown great potential in improving the performance of robot navigation. In response to the increasing deployments of mobile robots within various scenarios, a data-driven paradigm of navigation approach with safety verification is preferred where one can train RL algorithms with large amounts of prior data, keep learning continuously, and ensure safe navigation in applications. Conventional end-to-end reinforcement learning navigation paradigms have encountered multiple challenges in meeting these demands. In this work, we introduce a novel robot navigation approach termed NaviFormer. This approach handles navigation tasks based on sequence modeling to obtain the data-driven ability. It also integrates rule-based verification for safety insurance. We conduct a series of experiments to validate the data-driven ability of our approach and to compare it with existing navigation methods. We also perform quantitative tests on a real-world robot platform, TurtleBot. The experimental results show our method’s outstanding data-driven ability and highlight its superior arrival rate and generalization compared to other state-of-the-art methods like the PPO-based navigation method.