Invention Title:

CONTROL ARCHITECTURE FOR A MOBILE ROBOT

Publication number:

US20260326293

Publication date:
Section:

Performing operations; transporting

Class:

B25J9/1638

Inventors:

Assignee:

Applicant:

Smart overview of the Invention

The patent application describes a control system for mobile robots, focusing on dynamic balancing and manipulation capabilities. The system integrates multiple controllers and dynamics modules to manage the robot's movement and interaction with its environment. This architecture is particularly relevant for robots with articulated appendages, such as humanoid or quadruped robots, which require precise coordination for tasks involving both locomotion and object handling.

Control Architecture

The control system includes a first controller responsible for whole-body dynamics, ensuring the robot maintains balance and performs locomotion. A second controller manages the movement of specific robot parts, like end effectors, to interact with the environment. A dynamics module coordinates these controllers, allowing the robot to perform complex tasks that require both stability and dexterity.

Controller Functions

The first controller can be a model-based system, planning movements over a time horizon, or a learned policy-based system. The second controller also uses a learned policy, potentially trained through reinforcement learning, to control the robot's end effectors. These controllers operate with different time frames and sensory inputs, reflecting the distinct requirements of whole-body and end effector tasks.

Robot Capabilities

The described system allows a robot to dynamically balance while manipulating objects. The robot's articulated appendages, including legs and arms with end effectors like grippers, enable it to perform tasks such as grasping, manipulating, and releasing objects. The control system supports both single and bimanual interactions, enhancing the robot's versatility in various environments.

Advanced Features

The system includes features like impedance modules and reference trajectory information to refine control precision. Trajectory information involves position and velocity data, while learned policies provide stiffness and damping parameters. These elements contribute to the robot's ability to execute precise and adaptive movements, crucial for tasks requiring high levels of dexterity and coordination.