Control systems in a humanoid robot serve as the layer between the main AI system (that is responsible from the overalllearning, planning, reasoning and decision making of a robot) and the hardware that will enable robot’s actions.

According to the “order” given by the AI, the control system then executes this order, while at the same time coordinating all systems within the robot.

The highest control level is where the control system interacts with the AI. The middle control system controls overall posture and movement and the low level control systems controls joints (actuators) and sensors and processes sensor feedback. Let’s take a look at each.

High Level Controls:

High level controls interact with the AI to plan and coordinate mid level control tasks. The distinction between the AI and control system at this level is little. At this level, sensory inputs are interpreted, such as sensor fusion (combining data from multiple sensors to obtain a clearer picture), and based on these, and learned skills and memory, actions are planned, reasoning and decision making is performed, targets are set. For example a human may give command to a robot to take a box from here to the next room. It is at this level that the target is set and the overall planning for the required courses of action are made.

Middle Level Controls:

Mid level controls include balance and posture control, regulating stiffness and damping properties, gait and motion control, actual planning of trajectory. Multiple joints are coordinated to obtain a desired posture of a robot at a given instant, which may be a targeted posture or an intermediary one while executing some task while also remaining balanced. Concepts such as inverse kinematics is used to bring actuators in such a position to obtain a desired posture through coordinated limb movements. Middleware such as Robot Operating System (ROS), operates at this and higher level. Continuing from the example given above, after the robot sets its main target and understand what to do, the specific path to chose, and the necessary movements in general sense are planned at this level.

Low Level Controls:

Low level controls manage the actual physical components of the robot that will together execute the higher level decisions.

Actuation control: Precise control of motors, servos and any other type of actuators that there may be is essential for a desired robot operation. For example the position, velocity and torque of a servo is controlled at this level, by managing the supply of voltage and current and their timing to each component, in coordination with the overall higher level goals and other components. For example a command may be given to a joint such as “rotate joint 3 by 30 degrees at 8 degrees per second”. PID, torque and balance controllers are used for feedback. Stabilization and balancing algorithms are executed continuously during movement. Real Time Operating System (RTOS) operate heavily at this level. Continuing from the example above, at this level, according to the specific chosen path and planned body movements, actuators are instructed to perform in a specific way.

In all levels above, feedback loops are used to control and adjust behavior.

By: A. Tuter

——————————————————

Terms of use:

Copying or republishing of any type of content from this website is not allowed without written permission from us. We make dated records of our posts, keep originals of our images.

The content in this website may contain errors, inaccuracies or be incomplete. User assumes all liability as a result of usage of content in our site. Also see our Terms page for more details.