Hands On Robot Safety

Goals

The goal of this VIP is to build and control a mobile, bimanual manipulator robot that is capable of safely performing dexterous tasks in human home and work environments. Tasks include making hot tea, putting material in and out of a 3-D printer, and assembling furniture. The key is to create a unified hardware-and-software ecosystem that is both cheap and effective for solving complex tasks to help people.

Issues Involved or Addressed

Robots are rapidly becoming more and more common in human spaces, as we can see with autonomous mobile robots and vehicles around campus and in Atlanta. Soon, similar robots will enter our homes, workplaces, and common areas. However, it remains open what the best hardware and software options are for such systems; safer, softer hardware is more difficult to control and requires more complex algorithms, whereas more common (rigid) robots are inherently dangerous. PI Kousik’s team, the Safe Robotics Lab at Georgia Tech, has extensive experience designing algorithms for mathematically guaranteeing that robots will safe, and deploying software that ensures such algorithms are correctly implemented on real robot hardware. For this VIP team, we seek to answer the question, how can we safely have robots work near and with us? Towards solving this, we seek to build and control robots that can safely move around in human spaces and manipulate objects. Our team currently has a mobile manipulator system (see attached images, where we used Nano Banana to place our robot in realistic settings), but it is too large to use easily in human spaces. Thus, we are planning to build a smaller, more efficient system, with paired software and algorithms Potential research questions (social and technical) that we will address are:

• How can a robot efficiently generate safe motion to solve risky tasks, such as washing dishes in a kitchen or pouring molten metal in a forge?
• How can a robot concisely communicate in social interactions with humans and other robots to express its own safety and capabilities?
• How can a robot be used to create physical presence for dexterous, complex, remote manipulation?
• How can a robot safely and efficiently complete fast, high-risk tasks, such as hammering nails in a construction setting?