BEGIN:VCALENDAR VERSION:2.0 PRODID:Data::ICal 0.24 BEGIN:VEVENT DESCRIPTION: 'Title: AutoDriving CTF\n When: Friday\, Aug 12\, 09:00 - 08:59 PDT\n Where: Caesars Forum - Summit 206-208\, 238\, 237\, 234 (Con test Area) -\n [1]Map\n\n Description:\n Overview\n\n Last year\, we organized the AutoDriving CTF as an official contest of\n DEF CON 29 ([2]https://forum.defcon.org/node/237292) and did\n reasonably well: mor e than 100 teams participated and 93 teams had\n valid scores. Last year \, due to the pandemic\, the contest was online\n only with on-site demo nstrations. All the challenges were deployed in\n 3D simulators. This ye ar\, we propose a hybrid event with in-person\n challenges on-site. We a lso plan to introduce some new challenges with\n real vehicles involved\ , in addition to those based on autonomous\n driving simulators. We hope to continue the engagement with the\n hacking community to raise the aw areness of real-world security\n challenges in autonomous driving.\n\n The AutoDriving CTF contest focuses on the emerging security\n challeng es in autonomous driving systems. Various levels of\n self-driving funct ionalities\, such as AI-powered perception\, sensor\n fusion and route p lanning\, are entering the product portfolio of\n automobile companies. From the security perspective\, these AI-powered\n components not only c ontain common security problems such as memory\n safety bugs\, but also introduce new threats such as physical\n adversarial attacks and sensor manipulations. Two popular examples of\n physical adversarial attacks ar e camouflage stickers that interfere\n with vehicle detection systems\, and road graffitis that disturb lane\n keeping systems. The AI-powered n avigation and control relies on the\n fusion of multiple sensor inputs\, and many of the sensor inputs can be\n manipulated by malicious attacke rs. These manipulations combined with\n logical bugs in autonomous drivi ng systems pose severe threats to road\n safety.\n\n We design autonom ous driving CTF (AutoDriving CTF) contests around the\n security challen ges specific to these self-driving functions and\n components.\n\n The goals of the AutoDriving CTF are the followings:\n\n * Demonstrate se curity risks of poorly designed autonomous driving\n systems through hands-on challenges\, increase the awareness of\n such risks in sec urity professionals\, and encourage them to\n propose defense soluti ons and tools to detect such risks.\n\n * Provide CTF challenges that allow players to learn attack and\n defense practices related to aut onomous driving in a\n well-controlled\, repeatable\, and visible en vironment.\n\n * Build a set of vulnerable autonomous driving componen ts that can\n be used for security research and defense evaluation.\ n\n The contest is based on a Jeopardy style of CTF game with a set of\n independent challenges. A typical contest challenge includes a backend\ n that runs autonomous driving components in simulated or real\n envir onments\, and a frontend that interacts with the players. This\n year's contest will follow the style of last year and includes the\n following types of challenges: - “attack”: such as constructing\n adversarial patches and spoofing fake sensor inputs\, -\n “forensics”: such as i nvestigating a security incident related to\n autonomous driving\, - “ detection”: such as detecting spoofed\n sensor inputs and fake obstacl es\, - “crashme on road!”: such as\n creating dangerous traffic patt erns to expose logical errors in\n autonomous driving systems.\n\n Mos t of these challenges will be developed using game-engine based\n autono mous driving simulators\, such as CARLA and SVL. The following\n link co ntainssome challenge videos from AutoDriving CTF at DEF CON 29\n [3]http s://www.youtube.com/channel/UCPPsKbVpxwk-464KIzr8xKw\n\n What's new in 2 022\n\n This year\, we will unlock new security-critical driving scenari os such\n as stop-controlled and signalized intersections. New difficult y levels\n will be added to challenges in such scenarios by integrating real\n downstream AI modules such as object tracking from open-source\n autonomous driving software like Apollo\, Autoware and OpenPilot. For\n example\, players will be required to generate adversarial masks which\n will be overlayed on the surface of a stop sign to prevent the\n self -driving vehicle from stopping. The self-driving vehicle is\n equipped w ith a tracking component so merely hiding the stop sign in\n several fra mes will not work.\n\n A video demonstrating an attacked scenario is ava ilable at [4]https://youtu.be/4aedG1GNfRw\n\n In addition to the simulat ion challenges\, we will add challenges with\n real vehicles in the loop . In this setup\, the vehicle under attack\n will be placed on a rack an d the driving environment will be displayed\n on a monitor in front of t he windshield camera. We will have the real\n vehicle running in a lab a nd players and players will interact with\n the vehicle by remotely mani pulating the virtual surrounding\n environments (such as the projected r oad signs in front of the\n vehicle). The attack results will be judged based on systems logs (for\n open-source systems\, such as openpilot) or dashboard visualizations\n (for closed-source vehicles).\n\n The foll owing URL shows some specifications about the real vehicles\n [5]https:/ /docs.google.com/document/d/1oFC5Swn-UQ3hqIBA_Pw511o8WZqToU4TcQCb3UYocFc/e dit?usp=sharing\n\n In order to enable the audience to experience the ch allenges more\n directly\, we plan to set up a vehicle wheel controller on site this\n year. Audiences can drive themselves to compete with the self-driving\n vehicle in some of the challenges.\n\n For players\n\n * What do players need to do to participate AutoDriving CTF? Most of\n the challenges do not require domain knowledge of autonomous\n driving software or adversarial machine learning\, although\n know ledge of those helps. For example\, the players can generate\n image s the way they like (e.g.\, drawing\, photoshopping) to fool\n the A I-components or write a short python script to control the\n vehicle . Some challenges\, such as incident forensics likely would\n requir e players to learn domain knowledge such as sensor\n information for mat and how fusion works.\n\n * What do we expect players to learn thr ough the CTF event? Players\n can (1) gain a deep understanding of r eal-world autonomous driving\n systems' design\, implementation\, an d their corresponding security\n properties and characteristics\; an d (2) learn the attack and\n defense practices related to autonomous driving in a\n well-controlled\, repeatable\, visible\, and engagin g environment.\n\n Additional information\n\n Below are some materials from our first AutoDriving CTF at DEF CON 29\n in 2021\, which includes some challenge videos (Warning: the videos\n files could be large in go ogle drive)\, a summary of the event and some\n links reporting the even ts.\n\n [6]https://drive.google.com/drive/folders/1cr3qlX1mC7vGPzqqEZ900 ZDiEQdbzGo4?usp=sharing\n\n [7]http://www.buffalo.edu/ubnow/stories/2021 /11/team-cacti-capture-flag.html\n\n [8]https://medium.com/@asguard.rese arch/invisible-truck-gps-hacking-mad-racing-first-person-view-of-worlds-fi rst-ever-autonomous-9b2d5903672a\n\n [9]https://netsec.ccert.edu.cn/eng/ hacking/2021-08-06-autodrive-defcon\n\n [10]https://cactilab.github.io/c tf.html\n\n '\n\n 1. https://defcon.outel.org/consolidated_page.html#C aesarsSummitBR\n 2. https://forum.defcon.org/node/237292\n 3. https:// www.youtube.com/channel/UCPPsKbVpxwk-464KIzr8xKw\n 4. https://youtu.be/4 aedG1GNfRw\n 5. https://docs.google.com/document/d/1oFC5Swn-UQ3hqIBA_Pw5 11o8WZqToU4TcQCb3UYocFc/edit?usp=sharing\n 6. https://drive.google.com/d rive/folders/1cr3qlX1mC7vGPzqqEZ900ZDiEQdbzGo4?usp=sharing\n 7. http://w ww.buffalo.edu/ubnow/stories/2021/11/team-cacti-capture-flag.html\n 8. h ttps://medium.com/\n 9. https://netsec.ccert.edu.cn/eng/hacking/2021-08- 06-autodrive-defcon\n 10. https://cactilab.github.io/ctf.html\n\n\n DTEND:20220812T155900Z DTSTART:20220812T160000Z LOCATION:CON - Caesars Forum - Summit 206-208\, 238\, 237\, 234 (Contest Ar ea) SUMMARY:AutoDriving CTF END:VEVENT END:VCALENDAR