- ROS (Robot Operating System)
- An open-source framework for robot software development. Provides tools, libraries, and conventions for building robot applications. ROS 2 is the current version, used by most modern humanoid robots.
- SLAM (Simultaneous Localization and Mapping)
- Algorithm that allows a robot to build a map of its environment while simultaneously tracking its position within that map. Essential for autonomous navigation.
- Computer Vision
- AI that enables robots to interpret visual information from cameras. Includes object detection, face recognition, scene understanding, and visual navigation.
- Reinforcement Learning
- AI training method where a robot learns by trial and error, receiving rewards for successful actions. Used to teach robots locomotion, manipulation, and decision-making. Canada is a global leader (Amii, U of Alberta).
- Inverse Kinematics
- Mathematical process of calculating joint angles needed to place the robot's hand (or foot) at a specific position. Essential for reaching, grasping, and walking.
- Motion Planning
- Algorithms that compute a collision-free path from the robot's current position to a goal position. Must account for joint limits, obstacles, and dynamic environments.
- Teleoperation
- Remote control of a robot by a human operator. Many humanoid robots learn tasks through teleoperation first, then gradually become autonomous. Sanctuary AI uses this approach extensively.
- Digital Twin
- A virtual replica of a physical robot used for simulation, testing, and training. Allows developers to test code and behaviors before deploying on the real robot.
- Foundation Model
- Large AI model trained on diverse data that can be adapted for specific tasks. OpenAI and Google models are being applied to robotics, enabling robots to understand natural language and visual scenes.