UC Berkeley Capstone Project

Search & Rescue Robot

what is currently happening?

The urgency of the first 24 hours after an earthquake forces first responders to operate in unstable buildings during ongoing aftershocks, resulting in elevated injury risk.

65% of aftershocks happened within the first 24 hours of an earthquake

+

The survival rate of trapped survivor drops by 30% after the first 24 hours

=

46% of first responders are injured after a major earthquake

what does the survivors need?

What if there is a way to achieve all of the needs without the risks?

Big Impact in Small Packages

Our solution uses a compact robotic platform that can safely enter damaged buildings to support search and rescue operations. It can help locate survivors, communicate with them, and deliver basic first aid supplies, reducing the need for responders to enter hazardous environments before conditions are understood. The system can also provide information about interior conditions and survivor locations to support safer and more informed rescue decisions.

  • Pixelated map showing a navigation path with color-coded lines and arrows, indicating direction and movement.

    Mapping

    The LiDAR allows the system to understand the surrounding environment. With this information, the robot can determine the layout of the room as well as know where the robot is within that space.

  • A small robot with tank treads moving in a room with cardboard boxes, a tripod, a soccer ball, and a person partially visible.

    Navigation

    With mapping, the robot can patrol the area, navigating while updating its understanding of the environment. Using a wireless connection, operators can control the robot through a live camera footage to assess and guide next actions.

  • A small robot with wheels and electronic components on a wooden floor, with various cans and boxes in the background.

    Image Detection

    The front-facing camera is used to detect survivors, providing information about who they are and where they’re located. In our prototype, we use identifiable tags to simulate survivor detection and simplify image recognition.

  • Close-up of a section of a drone or robotic device with metal components, mounting screws, and wiring, on a white surface.

    Pick Up Tool

    Using a custom electromagnetic arm, the robot can carry aid supplies and transport them directly to survivors. For this proof of concept, magnetic attachments are used to allow simple and reliable pickup and release of the aid packages.

Full Robotic Software Architecture

ROS2 Foxy in Ubuntu Linux 20.04

Software Stack

ROS2 Jazzy

Ubuntu Linux

Navigation2

Cartographer SLAM

OpenCV

Python3

RDK X3

A* Pathfinding