• Skip to primary navigation
  • Skip to main content
  • Skip to primary sidebar
  • Skip to footer

Microcontroller Tips

Microcontroller engineering resources, new microcontroller products and electronics engineering news

  • Products
    • 8-bit
    • 16-bit
    • 32-bit
    • 64-bit
  • Applications
    • 5G
    • Automotive
    • Connectivity
    • Consumer Electronics
    • EV Engineering
    • Industrial
    • IoT
    • Medical
    • Security
    • Telecommunications
    • Wearables
    • Wireless
  • Learn
    • eBooks / Tech Tips
    • EE Training Days
    • FAQs
    • Learning Center
    • Tech Toolboxes
    • Webinars/Digital Events
  • Resources
    • Design Guide Library
    • DesignFast
    • LEAP Awards
    • Podcasts
    • White Papers
  • Videos
    • EE Videos & Interviews
    • Teardown Videos
  • EE Forums
    • EDABoard.com
    • Electro-Tech-Online.com
  • Engineering Training Days
  • Advertise
  • Subscribe

Researchers find a way to confuse autonomous vehicle cameras

August 8, 2017 By Lee Teschler Leave a Comment

A group of researchers have discovered that a few simple changes made to street signs can thoroughly bewilder machine learning algorithms that can ordinarily tell a stop sign from a speed limit sign. In a paper called Robust Physical-World Attacks on Machine Learning Models, researchers from the University of Washington, the University of Michigan, Stony Brook University, and UC Berkeley describe a program they created for testing purposes that mimics vandalism or graffiti on road signs. Showing the altered signs to autonomous driving systems resulted in a Stop sign being mistaken for a Speed Limit sign 100% of the time. In another case, the autonomous systems thought a Right Turn sign was either a Stop sign or an Added Lane sign.

The researchers say they are trying to come up with a better attack algorithm for testing the comprehension of machine learning systems applied to autonomous vehicles. Previous attack programs of this nature tend to generate unconvincing road sign camouflage that either isn’t realistic or doesn’t fool the machine learning algorithms used in autonomous vehicles.

right turn sign
Turn right? An autonomous vehicle may interpret this sign as a lane change indicator. Here a researcher holds the sign at slightly different angles to the camera.

It turns out that there are difficulties in developing a program that will reliably fool a sign recognition system. Researchers say physical attacks on a road sign must be able to survive such changing conditions as varying distances, angles, lighting, and the existence of debris. Additionally, vehicle cameras will not necessarily produce correctly scaled images as distances change. Adversarial perturbations in the image would need to survive such resolution changes and be correctly mapped to their corresponding physical locations.

Camouflage graffiti stickers tended to make image classifier algorithms interpret this sign as either a 45 mph speed limit or a yield sign.

Researchers used three different road sign modifications (subtle, camouflage graffiti, and camouflage art) that their attack algorithm (called RP2 generated. In poster-printing attacks, they printed a digitally perturbed true-sized image of either a Stop sign or a Right Turn sign, cut the print into the shape of the sign, and overlay it on a physical road sign. Subtle perturbations caused the Stop sign to be misclassified as a Speed Limit 45 sign, the misclassification target, in 100% of test cases. Poster-printed camouflage graffiti caused the Right Turn sign to be misclassified as a Stop sign, the misclassification target, 66.67% of the time. In sticker attacks, they printed the perturbations on paper, cut them out, and stuck them to a Stop sign. Sticker camouflage graffiti attacks caused the Stop sign to be misclassified as a Speed Limit 45 sign 66.67% of the time and sticker camouflage art attacks resulted in a 100% targeted misclassification rate.

camouflage art stop sign
Image classifier algorithms tended to see this sign with camouflage art as a 45 mph speed limit or a Lane Ends sign.

The point of all this, of course, was to come up with tougher test cases for the sign classifier algorithms now used in autonomous vehicles. Researchers say in future work, they plan to test their algorithm further by varying other conditions they haven’t yet considered, such as sign
occlusion.

You may also like:


  • Sensing is key to autonomous operation

  • ADAS developers contemplate sensor fusion
  • mentor centralized approach
    Centralized processing for autonomous vehicles
  • edge_case_predictions
    Simulations help explain why autonomous vehicles do stupid things

  • The danger of treating autonomous vehicle control as a solved…

Filed Under: Applications, Automotive

Reader Interactions

Leave a Reply Cancel reply

Your email address will not be published. Required fields are marked *

Primary Sidebar

Featured Contributions

Five challenges for developing next-generation ADAS and autonomous vehicles

Securing IoT devices against quantum computing risks

RISC-V implementation strategies for certification of safety-critical systems

What’s new with Matter: how Matter 1.4 is reshaping interoperability and energy management

Edge AI: Revolutionizing real-time data processing and automation

More Featured Contributions

EE TECH TOOLBOX

“ee
Tech Toolbox: Internet of Things
Explore practical strategies for minimizing attack surfaces, managing memory efficiently, and securing firmware. Download now to ensure your IoT implementations remain secure, efficient, and future-ready.

EE Learning Center

EE Learning Center

EE ENGINEERING TRAINING DAYS

engineering
“bills
“microcontroller
EXPAND YOUR KNOWLEDGE AND STAY CONNECTED
Get the latest info on technologies, tools and strategies for EE professionals.

RSS Current EDABoard.com discussions

  • Elektronik devre
  • Powering a USB hub: safely distributing current from a shared power supply
  • RF-DC rectifier impedance matching
  • How can I get the frequency please help!
  • 12VAC to 12VDC 5A on 250ft 12AWG

RSS Current Electro-Tech-Online.com Discussions

  • 100uF bypass Caps?
  • Fuel Auto Shutoff
  • Actin group needed for effective PCB software tutorials
  • how to work on pcbs that are thick
  • compatible eth ports for laptop

DesignFast

Design Fast Logo
Component Selection Made Simple.

Try it Today
design fast globle

Footer

Microcontroller Tips

EE World Online Network

  • 5G Technology World
  • EE World Online
  • Engineers Garage
  • Analog IC Tips
  • Battery Power Tips
  • Connector Tips
  • DesignFast
  • EDA Board Forums
  • Electro Tech Online Forums
  • EV Engineering
  • Power Electronic Tips
  • Sensor Tips
  • Test and Measurement Tips

Microcontroller Tips

  • Subscribe to our newsletter
  • Advertise with us
  • Contact us
  • About us

Copyright © 2025 · WTWH Media LLC and its licensors. All rights reserved.
The material on this site may not be reproduced, distributed, transmitted, cached or otherwise used, except with the prior written permission of WTWH Media.

Privacy Policy