Personal tools

AMZ Driverless Competition Embedded Systems Projects

From iis-projects

Revision as of 15:27, 30 October 2020 by Paulin (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search
AMZDriver.jpg

Short Description

AMZ driverless started in 2017 with the goal to design, transform and program a fast and reliable autonomous racecar. After winning the first ever Formula Student Driverless competition we started our second season willing to improve robustness and performance even further. We hope to be faster than a human driver in the coming years. You can see the results in the following video: https://youtu.be/FbKLE7uar9Y?t=35s

The teams are composed of ETH master students from diverse backgrounds. Joining the team implies working in an extremely fast-paced project with very smart and motivated engineers that share a passion, to push the limits of what has been done before. We partner with different labs at ETH to write semester theses related to the topics we investigate while working on the team.


The project will be done in collaboration with the AMZ (http://www.amzracing.ch/) based in Zurich.

The following projects are proposed:

Embedded stereo visual inertial system Standard commercial stereo visual inertial systems are rather limited on the market. They tend to have very small baseline or to be oversized and heavy. This project aims at creating a custom embedded system that precisely synchronizes 3 cameras and an IMU allowing running visual inertial algorithms in the main computing system. In addition, some image processing should be done at this system to reduce the overhead of the network and to compute the first steps on the computer vision detection algorithms.

Embedded machine learning for real time image processing Machine learning and especially deep learning algorithms are computationally expensive. This project aims at investigating HW accelerators for part of the object detection algorithms for computer vision and/or LiDAR.

Design and implementation of a rugged computing system for an autonomous racecar Autonomous vehicles require powerful computing systems but at the same time they should consume little power and be rugged and reliable. This project aims to design a custom computing system composed of off-the-shelf components and custom boards to efficiently combine the best of CPU and GPU under power and s restrictions.

Modular Battery Management System based on optical communication for an electric racecar AMZ electric developed an optical BMS for High Voltage accumulator that reduces the amount of wires, weight and allows for different packaging possibilities of the cells. This project aims at improving the flexibility of this system by exploring different communication options, and different HW-SW interfaces. FPGA’s look like a promising alternative to handle an “arbitrary” number of cells without running into runtime problems.

PCB and control logic design for safety critical actuators in an autonomous racecar Autonomous racecars have to be equipped with steering and brake actuators. These actuators are safety critical since AMZ racecars can reach from 0-100Km/h in under 2s. This project aims at investigating strategies to formally guarantee the correct behavior of these actuators. In addition they should be designed to transition to a safe state despite power loss, or other unforeseeable failure modes. These techniques should be non-programmable to avoid human errors and “bugs”.


Depending on the applicant's profile and project type, his tasks may involve some of the following:

  • lab. testing/characterization of RF localization modules and embedded systems: verification of the prototype's characteristics w.r. design specification (simulations), measuring power-consumption, and assessing detection performance in lab. conditions
  • High-level software programming, signal processing, machine learning, wireless communication
  • programming the circuit for a specific application, field testing, data acquisition
  • Machine Learning for microcontrollers or high performance embedde systems
  • PCB design to build a working prototype which includes all the subsystems


Status: Available

  • Looking for Semester and Master Project Students
Supervisors: Michele Magno;

Prerequisites

(not all need to be met by the single candidate)

  • Experience using the laboratory instrumentation - signal generators, oscilloscopes, DAQ cards, Matlab etc..
  • knowledge of microcontroller programming and PC programming (C/C++, preferably microcontroller with Bluetooth Low Energy but it is not mandatory)
  • basic knowledge or interests on signal processing, wireless communication for localization and machine learning is a plus
  • plus is also knowledge on printed circuit board (PCB) using Altium.


Character

20% Theory
50% Implementation
30% Testing and evalaution

IIS Professor

Luca Benini

↑ top

Detailed Task Description

Goals

Practical Details

Results