A group of scientists from Carnegie Mellon University (CMU) have developed robotic systems and planners to help robots explore unfamiliar environments more efficiently. The next generation of exploration robots can create more accurate and detailed maps. The CMU research team has been working on robotic exploration systems for over three years. The system can be connected to any robot and turn it into an autonomous explorer. The system consists of a computer connected to various sensors, including a 3D lidar sensor, a forward-facing camera, and inertial sensors. The robot is controlled by an exploration algorithm that allows it to know its location, where it has been, and where it should go next. The system works more efficiently than previous systems, creating real-time, detailed maps at twice the speed.
CMU’s exploration systems include three different modes. In the first mode, a human operator controls the robot’s movements while autonomous functions ensure safety. In the second mode, a human selects a point on a map, and the robot autonomously explores from there. In the third mode, the robot autonomously explores an entire room and creates a map of it. The CMU reconnaissance systems also work under difficult conditions with poor lighting or impaired communication. The systems have been tested with various robots and drones, mapping underground mines, a parking garage, and indoor/outdoor areas of the CMU campus. The system won an award in the Subterranean Challenge of DARPA, showcasing its flexibility for various applications such as delivery and search and rescue. The research team aims to share the results openly, enabling other scientists to build upon this system and develop more autonomous exploration robots for independent exploration.
