Modular, Easy to Build and Extend and Open Source Robotic Platforms

Source: Deep Learning on Medium

Modular, Easy to Build and Extend and Open Source Robotic Platforms

The world of robotics is no doubt thrilling. State-of-the-art robots in many different sectors including manufacturing and health have already shown a lot of promise. Advances in the field will, even more, transform the way we live and work. Robotics popularity is driven by the fact that they are becoming more efficient, precise, consistent and accurate compared to humans. Robotic research is also active with more researchers and engineers working towards achieving next-level robotics.

However, robotic research is not as cheap as it demands intensive computing power. Researchers are all the more working towards achieving low-cost robotic training, experimentation, and development.

Robotics Benchmarks for Learning with Low-Cost Robots

While the robotics industry is increasingly advancing, it is crucial for researchers to continuously work towards achieving low-cost robotic developments and solutions. A lot of such work has been going on such as the recent MuSHR RACE CAR Project.

In an effort to drive robotics research, yet another group of scholars recently released an open-source platform of cost-effective robots designed for reinforcement learning in the real world.

The platform introduces two robots to help accelerate reinforcement learning research in different task domains. One of the robots, D’Claw is a three-fingered hand robot that facilitates learning dexterous manipulation tasks, and the other D’Kitty is a four-legged robot that facilitates learning agile legged locomotion tasks.

D’Kitty tasks are centered around three commonly observed locomotion behaviors — Stand, Orient, and Walk. (Source: Google AI)
Co-ordinate — Policy trained via Hierarchical Sim2Real learns to coordinate two D’Kitties to push a heavy block towards a target (Source: Google AI).

Why is this Important?

Let’s look at the current situation: there are many advances and predictions in the robotic world. But the truth is we will keep seeing more jaw-dropping developments. Such will be driven by the availability of low-cost training and testing environments.

The ROBEL platforms help towards that. They are particularly modular with a significant cost advantage over other options which makes them ideal for scalable experimentation. Not only are they offering a low-cost solution, but they are also easy to maintain and robust enough to sustain on-hardware reinforcement learning from scratch.

“ ROBEL platforms are low cost, robust, reliable and are designed to accommodate the needs of the emerging learning-based paradigms that need scalability and resilience. We are proud to announce the release of ROBEL to the open-source community and are excited to learn about the diversity of research and experimentation they will enable.” — Google AI.

Code, documentation, results and more can be accessed here.

Read the full paper: Open-source, Cost-Effective Robotic Platform

Thanks for reading, please comment and share. For an update of the most recent and interesting research papers, subscribe to our weekly newsletter. You can also connect with me on Twitter, LinkedIn, and Facebook. Remember to 👏 if you enjoyed this article. Cheers!