// the take
A MATLAB object-oriented toolbox for serial-link manipulator kinematics/dynamics and mobile robot navigation (EKF, particle filter, D*, PRM, RRT), built as the companion code for Peter Corke's Robotics, Vision & Control textbook (1st/2nd editions). It's for MATLAB users teaching or learning classical robotics algorithms, not for anyone building a production robot stack.
The algorithms are old and well-exercised — this traces back to 1993 and has been a de facto reference implementation for serial-link kinematics/dynamics for two decades, so it's a solid thing to check your own implementation against. The SerialLink/Link object model maps cleanly onto DH/MDH parameters and covers forward/inverse kinematics, Jacobians, and full rigid-body dynamics (RNE) in one coherent API. Code generation (@CodeGenerator) can emit C, MEX, or Simulink blocks from the symbolic dynamics, which is a genuinely useful escape hatch given the interpreted M-code is slow. Mobile robotics coverage (Bug2, D*, PRM, RRT, EKF-SLAM, particle filter localization) is broader than most teaching toolboxes bother with.
It's explicitly dead — repostatus Inactive, the README's first line tells you to go use the Python successor instead, and issues go unanswered. Installation isn't self-contained: you need to separately clone and path two sibling repos (spatialmath-matlab, toolbox-common-matlab) or rely on the packaged .mltbx, which is a worse experience than a single `pip install`. It requires a MATLAB license, which is a real barrier compared to the actively maintained Python toolbox from the same author. The code trades performance for readability by design, so anything dynamics-heavy (RNE, trajectory optimization) will be noticeably slow versus a compiled or vectorized alternative, and the legacy Octave/ folder suggests Octave compatibility was bolted on and may have drifted.
We read the README, the directory tree and a key source file in October 2026. Nobody pays to be listed. How we judge →