Team Autonomous Robotics Competition Club (ARCC) from Penn State University
Team ARCC’s proposed solution was a platform (shown below)that performed search and rescue (SAR) missions, inspection and mapping missions, and other indoor tasks. The projected, overall system MSRP including the ground station, aircraft, avionics, and battery was $4,000. The platform was an X-configuration quadcopter that was small in size, allowing for operations in confined spaces. The selected configuration was a result of the design trade-off between endurance and maneuverability for indoor operations. The vehicle design was optimized to allow a high payload in terms of sensors that increase the versatility of the UAS by adding multiple features packed into a single solution. Appropriate selection of sensors allowed for a relatively low-weight system at 2.5 kg, which was comparable to other commercial solutions at the time.
The indoor navigation was based on a tracking camera that provided human pose information, allowing position hold flight mode. This feature was essential so that first responders can focus on identifying risks and locating survivors under collapsed structures instead of focusing on controlling the UAS. Additional range sensors in all four directions provided situational awareness for flying inside constricted space. A live video feed integrated with thermal images provided a complete visual solution to localize and detect people even in low light conditions; visible and infrared LEDs were used in similar situations. The visible LED was controlled by the operator while the infrared LED was automatically activated in low light conditions. Two-way radio communications provided an additional capability for first responders to communicate with any possible survivor. Propeller guards provided reliability for flying in an indoor environment.
Team ARCC’s vehicle for the FastFind mission was a LiPo battery-powered hexrotor capable of autonomous waypoint following. The system’s main priority was to support first responder teams conducting SAR missions. The main sensor was an integrated FLIR Thermal camera to assist in “FastFinding” and identifying missing persons in visually obscured scenarios. Real-time video transmission of the thermal camera feed was available to a ground operator. A trained object detection algorithm combined both RGB and thermal camera feeds to help identify potential missing persons during flights and to save geo-tagged images for processing offboard in case further analysis was required. Quick battery swaps allowed the vehicle to fly longer than the estimated 26 minute endurance. The system was low weight and low cost which supports the requirements for a first responder operation. It had a small footprint for easy transport and was capable of being deployed by a single operator.
Team ARCC’s vehicle for the LifeLink mission was a LiPo battery-powered hexrotor capable of autonomous waypoint following. The system’s main priority was to support first responder teams conducting missions where network connectivity is otherwise unavailable. They demonstrated an integrated AyrMesh hub which was capable of transmitting internet protocol-based data to multiple first responder groups to disseminate voice communications, images, and video to each group. Each ground hub could be placed up to 2 miles away from the drone-mounted hub, although this was not demonstrated, and provide up to a maximum bandwidth of 10 Mbps to nearby devices. Quick battery swaps allowed the vehicle to fly longer than the estimated 26 minute endurance. The system was low weight and low cost which supports the requirements for a first responder operation. It had a small footprint for easy transport and was capable of being deployed by a single operator.
Team ARCC’s solution focused on a scenario in which a public safety UAS is providing continuous broadband communications to first responders in a location without cellular coverage. Imagine a gradual spoof attack where fake GPS signals sent to the onboard GPS sensor cause the mobile station to drift far enough from its controlled hover position that communications become unavailable to the first responder team. The team’s countermeasure to this attack was to add a Visual Inertial Odometry (VIO) position data sensor to compare with GPS data for protection against GPS degradation or attacks on small, low cost UAS platforms. They modified the open-source PX4 flight control firmware to include the difference in GPS and VIO position data in the GPS quality checks. The autopilot switched to the backup position sensor when the GPS local position drifts far from the VIO sensor.
Vitor Valente is a Ph.D. student in Aerospace Engineering at the Pennsylvania State University. He earned both his B.S. degree in Control and Automation Engineering and M.S. in Mechanical Engineering from Federal University of Rio Grande do Sul, Brazil. His research interests include reconfigurable UAVs and adaptive control architectures. IFRS/Brazil, where he is currently a professor and has taught courses in Robotics and Pneumatics, is supporting his Ph.D. studies with the Penn State UAS Research Lab (PURL). He is a FAA certified remote pilot.
Rachel Axten is a NASA Pathways Intern and Ph.D. student in Aerospace Engineering at the Penn State UAS Research Lab (PURL). Her research interests include vehicle dynamics and fault-tolerant control. She is a FAA certified remote pilot and serves as the team's safety pilot.
Venkatakrishnan Iyer is a Ph.D. student in Aerospace Engineering at the Pennsylvania State University. He has a B.S. degree in Electrical Engineering from Mumbai University and a M.S. degree in Electrical Engineering from Penn State. His industry experience spans more than eight years, working at Hindustan Aeronautics Limited as a Systems Integration Designer for Helicopters. His current research involves dynamics and modeling of helicopters with development of adaptive control architectures and control allocation schemes.
Vidullan Surendran (team member in UAS 3.0 only) was a Ph.D. candidate in Aerospace Engineering at the Pennsylvania State University focusing on intent recognition in Human-Robot teams. He was an instructor for senior level courses on Advanced Programming Concepts in C++ and Software Engineering Techniques at the Penn State Aerospace Engineering department. As one of the founding members of ARCC, he participated in several vehicle platforms working on vehicle AI/decision making, software integration, electrical subsystem design and structural design.