Divyansh Gupta

Divyansh Gupta

Roboticist harmonizing controls and design

Blog posts

2024

Screw Theory: Modeling Manipulator

1 minute read

Published:

Screw theory is a mathematical framework to analyze and describe rigid body motion in 3D space. It elegantly connects the combined linear and rotational motion about same axis observed in a physical screw to rigid body motion. In addition, it gives better physical understanding of singularities compared to DH parametrization where the maths do not perfectly translate to geometric interpretation. In Screw theory, in case of singularities Screw Axis aligns or coincides with each other thus loosing a DOF.

2023

Reinforcement Learning: Introduction to RL with K-Bandit Problem

7 minute read

Published:

Reinforcement learning (RL) is learning what to do in different situations as to maximize a numerical reward. Now, what distinguishes RL from other types of learning is - learner is not told which actions to take, but instead learner must discover which action / series of actions will yield the most cumulative reward in some time-period. For example, a master chess player makes a move. The choice is informed by planning, anticipating possible moves and countermoves of the opponent and by judgement of the desirability of particular pieces on the board.