[robotics-worldwide] [software] Brittle Star Robot Simulator

Sunil Lal sunil at eva.ie.u-ryukyu.ac.jp
Tue Dec 29 12:42:52 PST 2015


Given a system consisting of homogenous sub-systems with a set of behavior and constraints, we were interested in knowing how the interaction of the sub-systems can lead to emergence of system level behavior. In particular we considered the brittle star typed robot as a highly nonlinear dynamic control system, which is composed of homogenous modules (sub-systems). We drew inspiration from emergent structures in nature and developed novel computational models for the motion control of the modular robot. Specifically, we leveraged off finite state machine, cellular automaton, and neural networks in developing novel motion control models. The optimal parameters for these models were obtained through evolutionary algorithms (GA & DE).

We are pleased to release the Brittle Star Robotic simulator, which was developed using open dynamics engine. It allows the user to experiment with different motion controllers, evolve motion control parameters, and visualize the robot's movement.

https://github.com/sunillal/robosim


Kind regards

Sunil Lal

Complex Systems Laboratory
Department of Information Engineering
University of the Ryukyus
Japan

On Nov 17, 2015, at 9:56 PM, Antonis Argyros <argyros at ics.forth.gr> wrote:

> 
> We are happy to announce the release of our new Hand Tracker library (http://cvrlcode.ics.forth.gr/handtracking/).
> 
> The new library brings a lot of bug fixes and updates, both under the hood and on the API level.
> 
> We are now providing fine grained control on the tracking pipeline through python, enabling easier modification and extension of the tracker for testing and academic use.
> 
> The new version is also fixing bugs that caused problems with some newer Nvidia drivers and GPUs.
> 
> The new library offers:
> - Depth based hand segmentation. No need for skin color.
> - Python API. Fine grained access to the tracking pipeline steps.
> - Tracking of two hands and hand-object tracking is now supported.
> - Improved performance (30+ fps) on modern GPUs.
> 
> Moreover, we are releasing the python scripts for this new version on github (https://github.com/FORTH-ModelBasedTracker/HandTracker). We plan to extend the number of provided tracking scripts over time and provide better support to the users of the library through the github tools.
> 
> Kind regards,
> 
> Antonis Argyros
> On behalf of the K3DHT team
> 
> 
> -- 
> Prof. Antonis Argyros
> Computer Science Department, University of Crete
> Tel: +30 2810 393594, argyros at csd.uoc.gr
> - and -
> Institute of Comp. Science - FORTH
> Tel: +30 2810 391704, argyros at ics.forth.gr
> http://www.ics.forth.gr/~argyros
> _______________________________________________
> robotics-worldwide mailing list
> robotics-worldwide at usc.edu
> http://duerer.usc.edu/mailman/listinfo.cgi/robotics-worldwide



More information about the robotics-worldwide mailing list