Collaboration Between Unmanned Aerial and Ground Vehicles Dr. Daisy Tang
Key Components Autonomous control of individual agent Collaborative system Mission planning Task allocation Communication Data fusion Operator control unit (OCU) Situation awareness Sliding autonomy
Paper Presentations Deploying air-ground multi-robot teams in urban environments, by Chaimowicz et al., Multi-Robot Systems, From Swarms to Intelligent Automata, 2005. Presented by Nada Alghofaili Integrated long-range UAV/UGV collaborative target tracking, by Moseley et al., SPIE Unmanned Systems and Technology XI Conference, 2009. Presented by Wie Li
Some Videos https://www.youtube.com/watch?v=fz AlGxrjg1g&spfreload=10 https://www.youtube.com/watch?v=r PCUB6xjQTI https://www.youtube.com/watch?v=el cctvwxddg
Motivation of Collaboration Challenges of urban environments Buildings pose 3-D constraints on visibility, communication and GPS A network of aerial and ground vehicles working in cooperation is more beneficial We need to: Keep the network tightly integrated for vehicles to support each other Provide ways to facilitate human operator to command the whole network
Research Goal Establish the overall paradigm, modeling framework and software architecture to enable a minimum number of human operators to manage a heterogeneous robotic team with varying degrees of autonomy
Hardware Team: 5 unmanned ground vehicles (UGVs) + 2 fixed wing aircraft and a blimp
UGVs 48 cm long and 35 cm high chassis on a scale model truck Pentium III laptop Odometry, steering servos GPS IMU A forward-looking stereo camera pair A small embedded computer with 802.11 wireless Ethernet Jbox handles multi-hop routing in an ad-hoc wireless network
UAVs Fixed wing aircraft: Equipped with Piccolo autopilot Provides innerloop attitude and velocity stabilization control A high resolution camera IMU GPS receiver Radio modem is used for communication between air vehicles and operator base station Blimp: 9 meters length, 3kg payload GPS, IMU, video camera, onboard computation and communication
Software ROCI (Remote Object Control Interface) for UAVs and UGVs A high-level OS for programming and managing networks of robots and sensors Each robot is a node that contains several processing and sensing modules and may export different types of services and data to other nodes Complex tasks can be built by connecting inputs and outputs of specific modules The connection is defined in XML
Localization and Navigation A Kalman filter is used to estimate robot localization based on Wheel encoder odometry, IMU, GPS, robot observations from external vision sensors and landmarks Navigation based on a list of waypoints Specified manually through a user interface Automatically generated Create a Voronoi Diagram of the environment and use it as a roadmap for planning intermediate waypoints Diagram can be generated beforehand using overhead imagery obtained by air vehicles Mission scripts
Trajectory Controller A trajectory controller generates linear and angular velocities Local obstacle avoidance is done by the two stereo cameras Trajectories can be compared to find potential collisions
Situation Awareness Main interface: ROCI Browser It displays the network hierarchically Human operator can browse nodes Tasks running on each node Modules that make up each task Browser s main job is to give user command and control over the network and ability to retrieve and visualize information from any one distributed node
Mission Scripts User can start and stop the execution of tasks in the robots remotely, change task parameters or control Elaborated missions are constructed using scripts, which define a sequence of actions that should be performed Capturing panoramic images at different waypoints, or navigating through multiple intermediate waypoints before reaching a target site
A Snapshot
Air-Ground Cooperation Challenges: cluttered urban environments UAVs could help UGVs by providing localization data and acting as communication relays Example: localize ground vehicles using a sequence of images taken from the blimp Relates the position of the robot in a global coordinate frame with its pixel coordinates in the image Use a set of known landmarks in the image Rely on measurements from the GPS/IMU onboard the blimp and camera parameters
Comparing Two Methods None of these approaches could be applied alone if we need a localization system that is applicable, reliable, and accurate Motivation: Find more sophisticated methods for cooperative localization Fuse information from different sources in a systematic way
The Combined Approach Based on prior work on decentralized data fusion (DDF) and decentralized active sensor networks A collaborative feature search and localization Exploits complementary character of UAV and UGV UAV rapidly covers designated search area UGVs deploy to refine the feature location
Example
Cooperative Radio-Mapping Communication is essential for coordination Radio propagation characteristics are difficult to predict Transmission power, terrain, 3-D geometry of the environment, interference Goal: Acquiring information for radio connectivity maps in urban terrains to help plan multi-robot tasks Approach: Build radio connectivity map, which returns signal strength between any two positions in the environment
Waypoints Navigation An overhead surveillance picture is used to generate roadmaps for motion planning and waypoints generation Minimize probability of losing connectivity under line-ofsight condition Radio signal strength measurements are obtained as team members simultaneously traverse through their respective waypoints Broadcast messages @ arrival Broadcast messages when ready to go after signal measurement Repeated until all waypoints are traversed Recovery behaviors: returning to the last position
Preliminary Results