🤖 STEAM x D 🤖
AI Rescue Operation Challenge 2023
Credit: midjourney, AI generated images.
Table of Contents
- Overview
- AI Rescue Operation Challenge
- General Rules
- Technical Rules
- Game Field
- Robot Assembly
- AI
- Raspberry Pi FPV Camera
Overview
A massive earthquake has struck the small country of Astoria. The epicenter of the earthquake is located at Rewa, a rural village best known for its holiday resort. Satellite images reveal that the village has been flattened by the earthquake, and the Rewa Holiday Resort and Spa has been completed destroyed. Hundreds of people are feared trapped and in need of rescue and medical aid. To compound this disaster, the President of Astoria was vacationing in Rewa when the earthquake occurred, and preliminary evidence suggests that she is most likely trapped somewhere amid the rubble. The village kindergarten is also located in the resort complex.
An AI-powered robot with image recognition technology has been deployed to Rewa before the arrival of search-and-rescue teams. One of the key tasks when entering a disaster area is to determine the search and rescue priority. And time is of the essence. Which specific area should be prioritized—and why? Who receives help first—and why? What are other factors that should be considered when strategizing this search-and-rescue operation? And might making one decision to help a particular party end up foreclosing the window to help another party? Because critical decisions that impact the well-being of survivors will have to be made, ethical reasoning is paramount.
As an engineer of an SUTD-based TECH NGO (a Start-up company), you have been contracted by the Astorian government to pilot this robot to assist with the search-and-rescue operation. Your TECH NGO is known for providing services that are technologically savvy and culturally competent, and such a human-centric approach to design is reflected in its work process. To manage the workflow, your organization has broken down the task into eight segments, which also the UK Design Council’s design thinking framework which is divided into the following phases: discovery, define, develop, deliver (see the table below for details). 
AI Rescue Operation Challenge
In this challenge, your team will need to combine all elements of STEAM x D to carry out the mission of rescuing victims from a disaster area using robots equipped with AI image recognition technology. Using state of the art AI-powered robots, your team will first need to modify and retrofit the robot to carry an onboard camera. This will be achieved using computer-aided design (CAD) and 3D printing technologies.
Your team will be tasked to explore a region of (120m x 50m). Given that it is a search and rescue mission, your team will have to navigate the terrain and successfully ‘rescue’ the injured/trapped ‘victim’/’pets’. A successful ‘rescue’ is accomplished when you’ve correctly identified the ‘civilians’/’pets’ for a minimum of 1 second.
In a single arena, a total of 15 minutes has been allocated for you to compete. You can attempt as many rescues as you need within the allocated time. Of all the attempts, the best attempt will be the score awarded to your team. Do note that another team will also be operating in the same 5 x 2.5m terrain area and you will need to communicate and negotiate the arena to ensure that no collisions take place.
There will be a total of 2 different arenas, outdoor and indoor, for which you will get 15 min on each terrain for your team to attempt. The overall score will be the sum of both mission scores.
Reference
ESP32Cam: https://github.com/arkhipenko/esp32-cam-mjpeg-multiclient
yolov5: https://github.com/ultralytics/yolov5
roboflow: https://roboflow.com/
Google's foundational course: https://developers.google.com/machine-learning/intro-to-ml
3Blue1Brown: YouTube Supplementary Notes