Search & Rescue

Robotic solutions that are properly sized with suitable sensors, mechanical structures, and navigation systems can improve the safety and security of first respondents as well as their work efficiency and flexibility. Based upon the advancements in automotive and aerospace technologies, the Search & Rescue project consortium designed and created an Obstacle Detection System (ODS), which was capable of fusing data collected by the system sensors and other external sources. Hence, the ODS was capable of providing first respondents with the full picture of the emergency environment, in which they were deployed.

Importance of Obstacle Detection System (ODS)

Worldwide teleoperated robots were successfully deployed in many rescue operations in urban environments with collapsed buildings and other structures [1]. From the after-action report’s analysis, one of the main areas of improvement was considered to be the enhancement of situation awareness of the robots [2]. Against this background an obstacle detection system was developed by the Search & Rescue project consortium. The teleoperated or semi-autonomous robots required a system for detecting moving objects as humans, vehicles and rescue dogs in the field. The system could support the operator or be used for obstacle avoidance during path planning from the robot.

The ODS was taking full advantage of all the available robot sensors to estimate with high confidence the object position. The obstacle detection system was also tracking the moving object’s and could further predict their future position, avoiding interfering with their motion. With the obstacle avoidance system, the robots could be safely used in the field and support the rescue units or move in areas dangerous for humans.

Advantages of multi sensor data fusion

Multi-sensors data fusion was used by the ODS to aggregate information coming from different sensors. The fusion is accomplished as a result of a Sensor Fusion Algorithm (SFA) which is executed after the object detection stage. A detection model was implemented and used for each type of input sensor. The input sensors used in Search & Rescue were the following:

• LiDAR sensor, for which a clustering algorithm was implemented to detect objects;
• Stereo camera, for which a Convolutional Neural Network was used to detect objects.

Once multiple objects were detected, the measures were aggregated and tracked. This was done with association algorithms (Global Nearest Neighbour) and tracking algorithms (Kalman filters).

Due to the Sensor Fusion Algorithm, more robust measures could be achieved, by employing filtering approaches. Moreover, the use of a sensor fusion algorithm allowed the robot to continue its task even in case of failure of a sensor (degraded mode).

Main objectives and functions

The main objectives of the ODS were the following:

• Detect and track obstacles, fusing information coming from different technologies of sensors;
• Notify detected objects to the robot. The robot would use this notification to stop itself in case of a potential collision;
• Make the robot’s pilot aware of the surrounding environment and notify him/her of obstacles detected;
• Cover different scenarios despite weather conditions and unavailable sensors.

Consequently, ODS main functions were:

• Acquire sensors (LiDAR and Camera) raw data;
• Filter and process raw data at sensors outputs;
• Detect obstacles in front of the robot, associating information from different sensors and tracking them

First aid device for kids

The second aim of the task was to design and develop an innovative protective first aid device for kids, which was able to carry the young victim safely out of the disaster scenario and monitor his health parameters. The device had also to help first responders to easily provide first aid support to injured children. The device had to be compatible with all the different paramedical devices (such as stretchers and spinal board) to immobilize the young victim, if necessary, before arriving at the hospital.

Health condition monitor for victims

The Emergency Response health condition Monitoring (ERM) device was a compact patient monitor that could display and transmit the health condition and position of the person wearing it through a smartphone. The device would be easy to handle and fast to apply without many cables interfering with the wearer’s motion with the purpose to be used on the field and track the condition of the victims.

Links

https://search-and-rescue.eu/

Literature

[1] Davids, A. (2002). Urban search and rescue robots: from tragedy to technology. IEEE Intelligent systems, 17(2), 81-83

[2] Murphy, R. R., & Burke, J. L. (2005, September). Up from the rubble: Lessons learned about HRI from search and rescue. In Proceedings of the Human Factors and Ergonomics Society Annual Meeting(Vol. 49, No. 3, pp. 437-441). Sage CA: Los Angeles, CA: SAGE Publications.

Keywords

Obstacle Detection System, multi sensor data, robot