A deep-learning point cloud network and sequence matching method are first integrated to detect loops, next the loops are closed by innovative algorithms and an Odometry module, thereby establishing a lightweight 3D LiDAR SLAM system as well as enabling mapping and navigating in large-scale, complex and unstructured scenarios. In addition, pose-graph optimization framework is adopted to tackle the configuration estimation issue. Obstacle-free and dynamically-feasible trajectories generated by our planner are proved to be highly accurate. With parameter adaptation, our controller is made compatible with heavy-loading and long industrial tractor-trailers, breaking prevailing technological constraints while maintaining outstanding control performance.

A novel quadruped robot with reformed motion/location control, compact size and joint motor driving algorithms is developed to move through different unstructured environments, including staircases, gullies and slopes. In addition, the brand-new joint actuator optimizes the stator, rotor as well as the embedded planetary gearbox with high power density. Don’t forget the permanent magnet synchronous motor with fractional slot concentrated winding, which promises low mass, large torque density, swift dynamic response and high load capacity. Kinematics and dynamics models as well as co-simulation scenarios are in place to better evaluate, validate and improve the motion control algorithms.

Capable of implementing SLAM by utilizing various revolutionary sensors in mapping unknown indoor environments, coupled with its collision avoidance feature, the newly developed Indoor Service Robot can navigate human indoor environments with ease. By enabling thorough evaluation of the environment, the robot gives assurance that it is compliant with high safety standards.

Daily tasks such as document delivery in offices can now be assigned to the Robot via the remote mobile app. Moreover, a remote server enables users to control the robot and observe its nearby environment remotely using the attached tablet, in which robotic sensors are installed for perceiving the surroundings.

The smart mobile base automatically aligns itself back to the charging pile when the battery runs out. This robot is readily applicable in indoor environments including but not limited to offices, hospitals, and libraries.

An AI-based inspection and monitoring system with multiple sensors, photo analytics, deep learning and pattern recognition techniques is to be developed for smart detection of building defects. To this end, first conduct aerial building inspection by UAV for Geo-ST data collection, DSM and DOM plotting with photogrammetry technology. Then, run the AI program to process the surveyed data and identify building defects. This unprecedented program revolutionizes the way of building inspection and its remote nature minimizes human interaction or impacts on nearby environments, which turn out to be all the more relevant and preferable in the wake of the global epidemic outbreak.