Multi-dimensional force sensors are widely installed in robotic arms. In industrial field production lines, multi-dimensional force sensors are mounted at the front end of a small robotic arm or at the end of a manipulator claw. Assist the robot arm to realize the functions of force control, contour tracking, hole search and anti-collision of the robot arm, so as to ensure the safety of robot operation and function realization.

Taking the control of mechanical arm as an example, each joint contains a clutch, brake and harmonic reducer. The motor is used as the power source, and the gear group and reducer provide kinetic energy for the joint. By adjusting the joint speed, position and force, the multi-degree-of-freedom rotational motion is completed.

2.2 Humanoid robots emphasize the “humanoid” attribute, and have higher requirements for gait control, impact resistance and trajectory planning, which significantly increases the difficulty

2.2.1 Lower limb control: It is difficult to control walking motion

Due to the humanoid leg structure, the motion control system of the humanoid robot in the walking state is nonlinear and strongly coupled. The humanoid robot needs to keep the walking stability and walk in accordance with the expected trajectory. At the same time, it is difficult to control the humanoid robot due to the interference of uneven ground and road obstacles. According to “Kinematic Analysis and Simulation of Humanoid Robot Based on Motion Capture Technology”, the lower limbs of humanoid robot can be simplified into a 14-degree of freedom system, in which the hip joint has 3 degrees of freedom, namely roll, pitch and deflection, and is connected by a Hooke pair and a rotating pair. The same transmission mode also acts on the 3 degrees of freedom of the ankle joint, 1 forward degree of freedom for each knee joint, connected by a rotating pair.

At present, the gait control method of humanoid robot is based on the control loop PID controller with feedback mechanism, which is optimized by PSO calculation. After optimization, the response speed of the control system and the improvement of the robot tracking path can be verified by Matlab simulation.

2.2.2 Arm control: visual feedforward + inverse kinematics solution to achieve trajectory planning, “human-like” attribute for impact and other indicators should be higher

Taking a four-DOF two-arm humanoid robot as an example, its motion control system consists of a robotic arm, a servo motor and a controller. The robotic arm contains two degrees of freedom in the shoulder and two degrees of freedom in the elbow.

The robot arm control is similar to the multi-joint control of industrial robots. The shoulder joint is taken as the origin of the coordinate system, and the end attitude of the robot arm and the positioning to be achieved are determined by machine vision, and then the analytical solution of the joint variables is obtained by solving the inverse kinematics algorithm. Finally, each joint is controlled to complete the task with a “human-like” attitude.

Jerk is the derivative of the acceleration speed during robot movement, which represents the speed of torque change. Jerk will cause problems such as vibration, overshoot, mechanical wear and reduced life. Considering that the robot arm of the humanoid robot has the “humanoid” attribute, it needs to grasp and lift items smoothly in operation, and it has higher requirements for achieving minimum impact.

2.2.3 Trajectory planning: Higher requirements are required for the integration, intelligence and visualization of trajectory planning algorithms

Humanoid robots to achieve “human-like” behavior, higher degrees of freedom than industrial robots, sensor applications will also be significantly increased, such as the need to introduce visual sensing to achieve interaction with the environment and spatial positioning (for trajectory planning). In industrial robot applications, the application of trajectory planning often requires professional engineers to process through programming, and the learning cost is high. Considering that humanoid robots have consumer application scenarios in the future, trajectory planning must be packaged through software, integrate functions and design a visual interface, so as to reduce the threshold of use.