RRO - Rapidly exploring Random Orienteering

Following video demonstrates the better runtimes of RRO over state of the art. State of the art takes over 150 seconds to solve a 10X10 grid. While we show that RRO finds a near-optimal solution within 12 seconds.






We demonstrate a challenging scenario for RRO in the following video, where the algorithm needs to sample specific nodes of high value to ensure it finds a near-optimal route.






Semantic Classification

The following video demonstrates our semantic classification algorithm. The images were collected during a manual flight test conducted at Fort IndianTown Gap. The classification algorithm can be run on-board the vehicle at 2 hz.






Gazebo/ROS based Simulator

We have developed a Gazebo/ROS based simulator that simulates the sensors on the vehicle with vehicle dynamics. Following video demonstrates exploration of an environment for cars through a UAV.





Autonomous Aerial Platform



We have performed succesful autonomous flights using our autonmous aerial platform. The vehicle is equipped with Lidar, a stereo pair, a high resolution camera and a GPS/INS unit. It is capable of autonomous waypoint following, semantic classification, visual odometry/SLAM and obstacle avoidance.