25 Non Trivial Robotics Concepts Every Robotics Aspirant Should Know About
From SLAM to Sim2Real.
Inverse Kinematics: Calculates joint parameters to place a robot end-effector at a target position and orientation.
Trajectory Planning: Determines the motion path of a robot considering velocity and acceleration constraints.
Kalman Filter: Estimates the state of a system using a series of noisy measurements.
Particle Filter: Implements recursive Bayesian estimation via Monte Carlo methods.
Visual Odometry: Estimates robot pose by analysing camera image sequences.
Simultaneous Localisation and Mapping: Builds a map of an environment while tracking the robot's position.
Probabilistic Roadmap: Constructs graphs of collision-free configurations for motion planning.
Rapidly Exploring Random Tree: Builds a tree to explore high-dimensional spaces for feasible paths.
Model Predictive Control: Optimises control inputs based on future state predictions.
Force-Torque Control: Regulates interaction forces during contact tasks.
Compliance Control: Adjusts motion based on force feedback for flexible interaction.
Redundancy Resolution: Resolves multiple solutions for over-actuated robots.
Dynamic Stability: Maintains robot balance during movement.
Zero Moment Point: Defines stable configurations in legged locomotion.
Whole-Body Control: Coordinates full robot motion across all joints.
Sensor Fusion: Integrates data from multiple sensors for improved state estimation.
Occupancy Grid Mapping: Uses probability grids to represent environments.
Markov Decision Process: Models decision-making in stochastic environments.
Partially Observable MDP: Extends MDPs to partially observable systems.
Reinforcement Learning: Trains agents via reward-based trial and error.
Deep Q-Network: Combines deep learning with Q-learning for high-dimensional problems.
Policy Gradient Methods: Optimises policies directly to maximise expected reward.
Transfer Learning: Applies pre-learned knowledge to new tasks.
Domain Randomisation: Improves generalisation by training across diverse simulated scenarios.
Sim-to-Real Transfer: Adapts simulation-trained models for real-world deployment.