Hackathon

This is a repo for the Hackathon Hack-the-car challenge. This project focuses on retrieving a 3d point cloud from the LiDAR sensor in order to predict bumps in front of the car.

Team MINOS

Members: Artur, Yang, Ishan, Ayush, Matteo

Table of Contents

1. Setup
2. Structure
3. Result

Setup

Needed python libraries:

• open3d
• pointcloud2
• PIL
• numpy
• ecal

Software & Interface

• eCAL
• Foxglove
• ecal-foxglove-bridge

How to run

• for offline mode, you need to prepare recorded data file, start eCAL player and play the data
• for online mode, connect your PC to device to get real time data
• start eCAL monitor, topic names and message types are presented
• start Foxglove and ecal-foxglove-bridge

python ecal_foxglove_bridge

• run protobuf_rec.py to subscribe data and detect barriers and bumps on the road surface

Structure

Communication

The communication interface and data flow is shown as below:

Algorithm

The following diagram describes the steps of retrieving obstacles (bumps) in front of the car. The algorithm is implemented in bump_detection.py.

Result

Foxglove visualization

ground points segmentation

point cloud clustering

bump detection