3D object detection combining semantic and geometric features from point clouds

Background: 3D object detection based on point clouds in road scenes has attracted much attention recently. The voxel-based methods voxelize the scene to regular grids, which can be processed with the advanced feature learning frameworks based on convolutional layers for semantic feature learning. The point-based methods can extract the geometric feature of the point due to the coordinate reservations. The combination of the two is effective for 3D object detection. However, the current methods use a voxel-based detection head with anchors for classification and localization. Although the preset anchors cover the entire scene, it is not suitable for detection tasks with larger scenes and multiple categories of objects, due to the limitation of the voxel size. Additionally, the misalignment between the predicted confidence and proposals in the Regions of the Interest (ROI) selection bring obstacles to 3D object detection. Methods: We investigate the combination of voxel-based methods and point-based methods for 3D object detection. Additionally, a voxel-to-point module that captures semantic and geometric features is proposed in the paper. The voxel-to-point module is conducive to the detection of small-size objects and avoids the presets of anchors in the inference stage. Moreover, a confidence adjustment module with the center-boundary-aware confidence attention is proposed to solve the misalignment between the predicted confidence and proposals in the regions of the interest selection. Results: The proposed method has achieved state-of-the-art results for 3D object detection in the Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) object detection dataset. Actually, as of September 19, 2021, our method ranked 1st in the 3D and Bird Eyes View (BEV) detection of cyclists tagged with difficulty level ‘easy’, and ranked 2nd in the 3D detection of cyclists tagged with ‘moderate’. Conclusions: We propose an end-to-end two-stage 3D object detector with voxel-to-point module and confidence adjustment module.

Kinematics analysis and gait planning for a hemiplegic exoskeleton robot

Background: As the world's aging population increases, the number of hemiplegic patients is increasing year by year. At present, in many countries with low medical level, there are not enough rehabilitation specialists. Due to the different condition of patients, the current rehabilitation training system cannot be applied to all patients. so that patients with hemiplegia cannot get effective rehabilitation training. Methods: Through a motion capture experiment, the mechanical design of the hip joint, knee joint and ankle joint was rationally optimized based on the movement data. Through the kinematic analysis of each joint of the hemiplegic exoskeleton robot, the kinematic relationship of each joint mechanism was obtained, and the kinematics analysis of the exoskeleton robot was performed using the Denavit-Hartenberg (D-H) method. The kinematics simulation of the robot was carried out in automatic dynamic analysis of mechanical systems (ADAMS), and the theoretical calculation results were compared with the simulation results to verify the correctness of the kinematics relationship. According to the exoskeleton kinematics model, a mirror teaching method of gait planning was proposed, allowing the affected leg to imitate the movement of the healthy leg with the help of an exoskeleton robot. Conclusions: A new hemiplegic exoskeleton robot designed by Shenzhen Institute of Advanced Technology (SIAT-H) is proposed, which is lightweight, modular and anthropomorphic. The kinematics of the robot have been analyzed, and a mirror training gait is proposed to enable the patient to form a natural walking posture. Finally, the wearable walking experiment further proves the feasibility of the structure and gait planning of the hemiplegic exoskeleton robot.

Development of dual-arm mobile robot platform based on ROS

Background: With the increasing demand of mobile robots in warehousing, logistics and service fields, simple planar motion is difficult to meet the task requirements of complex environment. The combination of mobile robot and cooperative robot is helpful to improve the dexterity of robot movement and expand the application of robots. Methods: Aiming at the application requirements of dual-arm robots and mobile robots in practical applications, this paper designed the hardware of a platform, built a simulation platform based on ROS (Robot Operating System), and designed the actual software control framework. Finally, the feasibility of the platform design was verified by the coupling motion experiment of the two robots. Results: We have established a simulation of the dual-arm mobile platform in ROS, designed the actual software control framework, and verified the feasibility of the platform design through experiments. Conclusions: The mobile platform can meet a variety of application requirements and lay the foundation for subsequent development.

A manipulator size optimization method based on dexterous workspace volume

Background: Manipulators for robots are required to have high manipulability for adaptability in different tasks. However, general methods for designing manipulators with high manipulability are deficient. Here, aiming at improving the manipulability of the six degrees-of-freedom (DOF) manipulator, a method for optimizing structure size parameters based on the dexterous workspace volume is proposed. Methods: Firstly, the kinematic analysis of the manipulator is performed. Then, the manipulability of the single working point of the manipulator is judged based on reachability in different postures. The workspace of the manipulator is discretized to obtain the volume of the overall dexterous workspace. By taking the maximum volume of the dexterous workspace as the optimization goal, Genetic Algorithm (GA) is used to optimize the structure size parameters to achieve optimal manipulability. AUBO 6-DOF manipulator is defined and analyzed as an example. Results: The optimization results indicate that the dexterous workspace of the manipulator expands in volume and its manipulability is improved. The validity of the proposed optimization method is verified by comparing the distribution of dexterous workspace of the manipulator. Conclusions: This article proposes an optimization method for the structure size parameters of a 6-DOF manipulator, which can be implemented to improving the manipulability of the manipulator.