Accurate inertial parameter identification is crucial for the simulation and control of robots encountering intermittent contacts with the environment. Classically, robots' inertial parameters are obtained from CAD models that are not precise (and sometimes not available, e.g., Spot from Boston Dynamics), hence requiring identification. To do that, existing methods require access to contact force measurement, a modality not present in modern quadruped and humanoid robots. This paper presents an alternative technique that utilizes joint current/torque measurements -a standard sensing modality in modern robots- to identify inertial parameters without requiring direct contact force measurements. By projecting the whole-body dynamics into the null space of contact constraints, we eliminate the dependency on contact forces and reformulate the identification problem as a linear matrix inequality that can handle physical and geometrical constraints. We compare our proposed method against a common black-box identification method using a deep neural network and show that incorporating physical consistency significantly improves the sample efficiency and generalizability of the model. Finally, we validate our method on the Spot quadruped robot across various locomotion tasks, showcasing its accuracy and generalizability in real-world scenarios over different gaits.

Autonomous assembly is an essential capability for industrial and service robots, with Peg-in-Hole (PiH) insertion being one of the core tasks. However, PiH assembly in unknown environments is still challenging due to uncertainty in task parameters, such as the hole position and orientation, resulting from sensor noise. Although context-based meta reinforcement learning (RL) methods have been previously presented to adapt to unknown task parameters in PiH assembly tasks, the performance depends on a sample-inefficient procedure or human demonstrations. Thus, to enhance the applicability of meta RL in real-world PiH assembly tasks, we propose to train the agent to use information from the robot’s forward kinematics and an uncalibrated camera. Furthermore, we improve the applicability by efficiently adapting the meta-trained agent to use data from force/torque sensor. Finally, we propose an adaptation procedure for out-of-distribution tasks whose parameters are different from the training tasks. Experiments on simulated and real robots prove that our modifications enhance the sample efficiency during meta training, real-world adaptation performance, and generalization of the context-based meta RL agent in PiH assembly tasks compared to previous approaches.

Existing view planning systems either adopt an iterative paradigm using next-best views (NBV) or a one-shot pipeline relying on the set-covering view-planning (SCVP) network. However, neither of these methods can concurrently guarantee both high-quality and high-efficiency reconstruction of 3-D unknown objects. To tackle this challenge, we introduce a crucial hypothesis: with the availability of more information about the unknown object, the prediction quality of the SCVP network improves. There are two ways to provide extra information: first, leveraging perception data obtained from NBVs, and second, training on an expanded dataset of multiview inputs. In this work, we introduce a novel combined pipeline that incorporates a single NBV before activating the proposed multiview-activated (MA-)SCVP network. The MA-SCVP is trained on a multiview dataset generated by our long-tail sampling method, which addresses the issue of unbalanced multiview inputs and enhances the network performance. Extensive simulated experiments substantiate that our system demonstrates a significant surface coverage increase and a substantial 45% reduction in movement cost compared to state-of-the-art systems. Real-world experiments justify the capability of our system for generalization and deployment.

In this paper, we introduce a novel approach to centroidal state estimation, which plays a crucial role in predictive model-based control strategies for dynamic legged locomotion. Our approach uses the Koopman operator theory to transform the robot’s complex nonlinear dynamics into a linear system, by employing dynamic mode decomposition and deep learning for model construction. We evaluate both models on their linearization accuracy and capability to capture both fast and slow dynamic system responses. We then select the most suitable model for estimation purposes, and integrate it within a moving horizon estimator. This estimator is formulated as a convex quadratic program to facilitate robust, real-time centroidal state estimation. Through extensive simulation experiments on a quadruped robot executing various dynamic gaits, our data-driven framework outperforms conventional Extended Kalman Filtering technique based on nonlinear dynamics. Our estimator addresses challenges posed by force/torque measurement noise in highly dynamic motions and accurately recovers the centroidal states, demonstrating the adaptability and effectiveness of the Koopman-based linear representation for complex locomotive behaviors. Importantly, our model based on dynamic mode decomposition, trained with two locomotion patterns (trot and jump), successfully estimates the centroidal states for a different motion (bound) without retraining.

Reinforcement learning (RL) has recently proven great success in various domains. Yet, the design of the reward function requires detailed domain expertise and tedious fine-tuning to ensure that agents are able to learn the desired behaviour. Using a sparse reward conveniently mitigates these challenges. However, the sparse reward represents a challenge on its own, often resulting in unsuccessful training of the agent. In this paper, we therefore address the sparse reward problem in RL. Our goal is to find an effective alternative to reward shaping, without using costly human demonstrations, that would also be applicable to a wide range of domains. Hence, we propose to use model predictive control (MPC) as an experience source for training RL agents in sparse reward environments. Without the need for reward shaping, we successfully apply our approach in the field of mobile robot navigation both in simulation and real-world experiments with a Kuboki Turtlebot 2. We furthermore demonstrate great improvement over pure RL algorithms in terms of success rate as well as number of collisions and timeouts. Our experiments show that MPC as an experience source improves the agent's learning process for a given task in the case of sparse rewards.

Viewpoint planning is an important task in any application where objects or scenes need to be viewed from different angles to achieve sufficient coverage. The mapping of confined spaces such as shelves is an especially challenging task since objects occlude each other and the scene can only be observed from the front, posing limitations on the possible viewpoints. In this paper, we propose a deep reinforcement learning framework that generates promising views aiming at reducing the map entropy. Additionally, the pipeline extends standard viewpoint planning by predicting adequate minimally invasive push actions to uncover occluded objects and increase the visible space. Using a 2.5D occupancy height map as state representation that can be efficiently updated, our system decides whether to plan a new viewpoint or perform a push. To learn feasible pushes, we use a neural network to sample push candidates on the map based on training data provided by human experts. As simulated and real-world experimental results with a robotic arm show, our system is able to significantly increase the mapped space compared to different baselines, while the executed push actions highly benefit the viewpoint planner with only minor changes to the object configuration.