Project Name: RoboCup Rescue League Robot

Developer: Rescube Robotics

DescriptionAdaptive custom mobility robot that can sense its environment and find earthquake victims via Flir Lepton Thermal Imaging, CO2 breath sensors and movement detection

Status: Prototype (used in competitions)

Date submitted to FLIR: Oct 30, 2016

About the Project:

The team of friends at Rescube built their robot for the RoboCup Major Rescue League, a competition featuring robots traveling through very difficult terrains that also gather and process the most information possible to detect and localize victims and then help the rescue personnel to ensure their safety.  The robot, an adaptive custom mobility platform, has some great tricks under the hood like the wide wheels with thin spine between, to never get stuck with the underside of the robot. The wheels incorporate critical parts like batteries, motor controllers, motors, gearboxes, and some extra weight to make to robot more stable. This configuration also lets them change the wheelbase depending on the situation, as a shorter robot can move better in confined spaces, but a longer one can climb steeper ramps and stairs. The elastic parts on the wheels handle the conflicts between the terrain and the robots movement, and also the spine can twist, so all the wheels touch the ground all the time. The robot has quite a lot peripherals attached to their two onboard computers (running linux & ROS) including:

  • (2) microcontrollers help the high level systems get down to the analog-digital or timing sensitive protocols, like generating PWMs for the 13 servos (some seriously altered mechanically and electronically), and interfacing with the gyroscopes, and other sensors
  • (6) RGB cameras 
  • (2) Flir Lepton thermal imaging sensors
  • Structure.io depth sensor
  • CO2 sensor for sensing breath
  • 2d stabilised Lidar, for Simultaneous Localization and Mapping (SLAM)
  • Microphone
  • Speaker
  • 9d IMU (gyroscope, accelerometer, magnetometer)
  • Barometer (for calculating altitude changes based on air pressure)

 For best connection they use the standard Lepton UVC interface, and use the OpenCV-based cv_camera ros node on their Odroid XU4 SBC.  For the most reliable connection they opted for some serious network gear, made by Mikrotik, and used directed (sector) antennas to ensure great signal-to-noise ratio. Their setup connects on 5ghz WiFi by default, but when that link breaks (it does not like walls much) they fall back to 2.4ghz, and when that fails they still have 4g LTE connection between the base and the robot.   The robot needs to manipulate the environment doing tasks like turning valves, disconnecting electric devices, moving objects to build supporting structures, placing sensors in the environment, and giving water/food/handheld radios to the victims found. To support dexterity tasks like this they custom built a robotic arm, and some external mechanics including:

  • Custom made robotic arm with 4 degrees of freedom
  • (2) grippers, with 3 degrees of freedom
  • (2) pan-tilt camera bases with 3 cameras on each

 The team achieved several important awards at RoboCup WorldCup 2016 including:

  • Major Rescue League 9th place
  • Best Outdoor CarryBot Award
  • RoboCup Design Award Sponsored By Flower Robotics

 You can follow Team Rescube’s progress for the upcoming RoboCup competitions at http://rescube.hu/blog and https://www.facebook.com/RescubeRobotics.