At the moment it is the vision of Real-World ML to develop a platform upon which different algorithms can be tested and ultimately compared. Towards that end an ecosystem should be put in place with a number of base plates and multiple community-sourced extensions. Furthermore a Simulation of these should be provided for training purposes. All parts should be open-source and cheaply build-able for people with access to basic materials and a 3D printer.
Currently our focus lies on providing a six legged robot that can be used to train walking algorithms. Towards that end a cheap first platform should be launched and integrated into the current ecosystem of Agents, with Simulation based upon the platform of Unity ML-Agents.