Achieving precise trajectory tracking for autonomous mobile robots in complex and dynamic environments pose a demanding challenge. In this study, we propose an innovative approach for the online refinement of model predictive control (MPC) through the application of double Q-learning, designated DQMPC. This method harnesses the dynamic interaction capabilities of double Q-learning with operational environment, facilitating the adaptive tuning of MPC parameters to improve the control performance. To enhance the computational real-time performance of the double Q-learning method, we develop an incremental discretization approach that performs nonuniform discretization of the action and state spaces to improve the learning efficiency. Additionally,We use a time-error-based prioritized experience sampling method to reduce the interdependence between past experiences and thus accelerate the training speed. Through extensive experiments, we validate the effectiveness of our DQMPC method which consistently outperforms traditional control technologies.

Direct estimation of body weight through noncontact methods is crucial for applications such as health monitoring, surveillance, and robot-assisted casualty rescue. Existing methods for body weight estimation from images are often affected by various factors, such as camera distance, human orientation, and body posture, etc., which ignore the fact that bodies typically inhabit 3D space. To address the problem, we propose an approach for estimating body weight based on virtual anthropometric measurements and deep features instead of estimating by reasoning pixels. Specifically, we develop a three-branch framework that includes face feature extraction, body feature extraction as well as deep feature extraction, and maps all features with a regressor. The developed method adds 3D shape reconstruction to explicitly reason about virtual anthropometric measurements. To enable this, our model is trained to robustly compute anthropometric measurements in various orientations and postures. Furthermore, we evaluate our method on a public dataset and Image-VM-BMI, a new dataset of 4740 images, including BMI labels and virtual anthropometric measurement labels with paired 3D reconstruction. Extensive experimental results demonstrate that the proposed method outperforms pixel-based analysis approaches on BMI estimation.

When disasters such as earthquakes and collapses occur, rescuers must go deep into dangerous regions and rush to transport the wounded. Long-term rescue operations in dangerous environments challenge the rescuers’ physical strength and threaten their lives. We propose a human-body reconstruction method for autonomous rescue operations with a wounded rescue robot. This method constructs a human body model that can support a robot’s autonomous rescue operations given limited scans of the injured. Most casualties lie on the ground in supine positions, so this scanning limit is typical. We propose a scanning viewpoint planning method using deep reinforcement learning and change the viewpoint with the help of a 6-degree-of-freedom (DOF) robotic arm to obtain a complete scan of the visible surface. For successful autonomous casualty rescue, a robot must have access to a watertight human body mesh with precise surfaces and associated semantics. To this end, we develop a neural network that predicts a completely clothed body mesh from visible point clouds. The results of experiments show that our method can reconstruct a watertight body mesh with accurate surfaces and human semantics to satisfy the needs of autonomous robotic casualty rescue operations under scanning limitations.

Most vision-based robotic tracking methods focus on general objects instead of human beings. This limitation stems from the inherent morphological variations in humans, characterized by non-rigid deformations and diverse kinematic configurations, which constrains the reliable limb tracking through conventional methods. The 3D representation of the human body provides rich spatial information and the inherent semantics of the limbs offer stable target signatures, while these properties remain under-exploited in dynamic motion tracking tasks. To address these challenges, we propose a dual-view fusion of perception-to-action framework for limb tracking, incorporating cross-view feature aggregation. This dual-view visual fusion attention mechanism captures both body shape and posture variations, as well as maintaining egocentric-perception of the target limb. The reasoning and planning capabilities across perspectives also be enhanced through this approach. Additionally, we present a feature extractor based on human semantic parsing to improve the limb perception. To further reduce the complexity of human image features and increase sim-to-real transfer efficiency, we develop an image matching mechanism based on 3D human reconstruction. Through the examination of various simulated and real-robot examples, our methods exhibit superior properties compared with current vision-based methods in terms of the success rate.