Navigation

The 2016 AMOS Project Show Case: CAODIC

Team / Project CAODIC reports:


CAODIC – Collision Avoidance using Object Detection and Inter-car Communication

As part of the course AMOS (Agile Methoden und Open Source), Team 5 developed a program for object detection and inter-car communication.

Summary The project implements a scenario needed for autonomous driving with the help of software agents: A bus stops at a bus stop vis-à-vis to a school. A car that drives behind the bus cannot see the children/humans starting to run across the street in front of the bus. Another car on the other side of the street detects the children and is able to warn the first car.
Overview
Main tasks
  • Integration and implementation of a simple bus/car and human detection algorithm
  • Integration of a software agent library
  • Implementation of the scenario using software agents and inter-process communication
Details The project involved developing a prototype that demos two communicating, virtualized hosts. Each host represents a car and one host (the client) warns the other (the server) about humans standing in front of a car/bus. The program reads and decompresses the video data and performs the {car,human}-detection for each frame. If the program detects a human standing in front of a car, a warning message is sent from the client to the server.
Setup
  • Operating System: Linux Ubuntu 14.04
  • SW Agents Library: C++ Actor Framework (CAF)
  • Object Detection Library: OpenCV
  • Messaging Format: HDF5, Google Protobuf
  • Deployment: Docker, Travis and Jenkins
  • Virtualization: Docker
Industry partner Continental AG, Regensburg
Coach Prof. Dr. Dirk Riehle
Team members Richard Fuchs, Daniel Götz, Nils Häusler, Jonas Heinrich, Elisabeth Hoppe, Leonard Keidel, Debin Liu
Outcome The following video shows the outcome of our project, including the {car,human}-detection, as well as the communication:

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.