OpenBot: Turning Smartphones into Robots

Regardless of recent innovations in the field of robotics, most robots are possibly pricey or have really confined features. On the other hand, a important part of the entire world populace individual smartphones, which are geared up with impressive CPUs, cameras, GPS, Wi-Fi and other technologies. A the latest research implies using a smartphone for computation and sensing of a robot.

OpenBot: method overview. Graphic credit history: Matthias Müller, Vladlen Koltun, arXiv:2008.10631

The authors of the research propose an electric car or truck which is in a position to perform simple responsibilities this kind of as man or woman next or authentic-time autonomous navigation. The robot can navigate alongside a new route and recognise new obstacle configurations not incorporated in the instruction. It can also recognise dynamic hurdles and stand up to motion blur because of to vibrations. The machines relies on readily readily available electronics (like Arduino Nano board) and 3D printing for its chassis. A single body for the robot prices $fifty and the rate can be lessened in the case of mass generation.

Present robots are possibly pricey or make important compromises on sensory richness, computational ability, and interaction abilities. We propose to leverage smartphones to equip robots with comprehensive sensor suites, impressive computational capabilities, condition-of-the-art interaction channels, and accessibility to a flourishing program ecosystem. We style and design a little electric car or truck that prices $fifty and serves as a robot body for normal Android smartphones. We build a program stack that permits smartphones to use this body for mobile procedure and display that the method is sufficiently impressive to help advanced robotics workloads this kind of as man or woman next and authentic-time autonomous navigation in unstructured environments. Controlled experiments display that the presented strategy is sturdy across distinct smartphones and robot bodies. A video of our perform is readily available at this https URL

Link: muscles/2008.10631