Code & Data

Code:

Fruit Gym

Fruit Gym is a collection of Gymnasium environments for robotic manipulation tasks—specifically, for picking strawberries using a Franka Panda robot. These environments support domain randomization, image observations, and multi-target grasping. They are designed to challenge reinforcement learning algorithms in continuous control and multi-object manipulation tasks.

The code can be found here and publised in: [1]

SplatGym

SplatGym is a simulator for reinforcement learning free space navigation policies in Gaussian splat environments.

It has the following main features:

  • Novel View Synthesis - Generate photo-realistic images of the scene from any arbitrary camera pose.
  • Collision Detection - Detect collision between the camera and the underlying scene objects.

The code can be found here and publised in: [2]

URmat

The URmat project is an interface between Matlab and Universal Robots (UR) robotic manipulators. It enables the control of UR robots with Matlab programming over TCP/IP connection. It can be useful for teaching robotics and implementing simple projects. Also URmat can be used alongside with Peter Corke’s Robotics Toolbox. It has been developed and tested on a UR5 robot with software version 3.0

The code can be found here and comes without any guarantee. Developers and testers are welcome!


Data:

Inverse Dynamics Datasets:

  • BaxterRhythmic – dynamics generated by a circular movement of the Rethink Baxter robot
  • URpickNplace – dynamics generated by pick n place of a 5 kg object for the Universal Robots UR10 manipulator
  • BaxterRand – dynamics generated by point reaching movements of the Rethink Baxter robot

The datasets can be downloaded here and made availiale in: [3]

Forward Dynamics Datasets:

  • KukaDirectDynamics – 10 trajectories generated from Kuka LWR while executing a pick and place task
  • BaxterDirectDynamics – 10 trajectories generated from Rethink Baxter while executing a pick and place task

The datasets can be downloaded here and made availiale in: [4]

References:

  • Zero-Shot Sim-to-Real Reinforcement Learning for Fruit Harvesting
    By E. Williams and A. Polydoros
    In 2025 IEEE 21th International Conference on Automation Science and Engineering (CASE)
    2025.
    Details
  • Robotic Learning in your Backyard: A Neural Simulator from Open Source Components
    By L. Zhou, O. Sinavski, and A. Polydoros
    In 2024 Eighth IEEE International Conference on Robotic Computing (IRC)
    pp. pp. 131–138, 2024.
    Details
  • A reservoir computing approach for learning forward dynamics of industrial manipulators
    By A. S. Polydoros and L. Nalpantidis
    In 2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
    pp. pp. 612–618, 2016.
    Details
  • Real-time deep learning of robotic manipulator inverse dynamics
    By A. S. Polydoros, L. Nalpantidis, and V. Kruger
    In 2015 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)
    pp. pp. 3442–3448, 2015.
    Details