When and Where
Wednesdays, 4pm–6pm (when there is no conflicting colloquium) or 5pm–7pm (when there is), in FB 007.
What and How
This course is being organized by three faculty members: Jim Anderson, Sanjoy Baruah, and Alex Berg. The focus of the course is automotive cyber-physical systems (CPSs). CPSs are systems that have designs that reflect a coupling of computational and physical elements. In the automotive case, we are mostly interested in new features that are making vehicles more and more autonomous.
Many CPSs must operate in situations where accurately understanding and predicting environmental conditions is essential. To provide such environmental awareness, advanced sensor technology is required that can generate significant amounts of data that must be processed in real time. The resulting data-processing rates can be difficult to sustain computationally. One potential solution is to resort to hardware over-provisioning. However, this is a wasteful practice that can be untenable in many domains due to monetary cost restrictions and size, weight, and power (SWaP) limitations. The development of system design and analysis methods that yield fielded products that meet these restrictions and limitations remains an important challenge.
In the automotive domain, a proliferation of advanced sensor technology is being fueled by an expanding range of autonomous capabilities. Driver-assist features, such as blind spot warnings, automatic lane-keeping, adaptive cruise control, and collision avoidance/mitigation systems, are becoming commonplace in high-end vehicles; in the coming years, such features are expected to evolve to provide significantly enhanced functionality, such as pedestrian detection, cross-traffic alerts, traffic sign recognition, and 360-degree sensing. At the same time, fully autonomous vehicles have been demonstrated in various "one-off" settings; these include the press-worthy Google Car, and various research vehicles that were fielded as part of the DARPA Urban Grand Challenge.
These autonomous features leverage work within an intersection of areas, including (of relevance to this department) real-time and embedded systems, computer security, computer vision, multicore and GPU computing, operating systems, and robotics. The goals of this course are threefold: (i) to bring together people working in these areas, (ii) to give them a forum for learning about each others' research areas and specific research activities in the department that might be applicable to automotive use cases, and (iii) to give them a forum for learning about automotive features where enhanced autonomy and situational awareness are required as envisioned within the automotive industry.
Procedurally, this will actually be a student-led course. We will meet once a week (each Wednesday), and each week, one or more students will be responsible for presenting that day's material. Ideally, we would like to have about 20 students in the class, and ideally, we would like to draw students from each of the background areas mentioned in the prior paragraph. Our current plan is to break students into teams (of say 4 students each), break the material to be covered into several distinct topics (let's say 5 such topics, for the sake of discussion), and then require each team to present enough material to cover some fraction of the classes (given these numbers, that would be around 3 classes per team). We will have to work out these details exactly after we know who has registered. Additionally, this course was partially motivated by a new NSF project, involving researchers in the department and at General Motors, on the realization of new automotive features where sensing is done via cameras, and vision-based processing streams must be supported in real time. There is a possibility that researchers from GM may be willing to give a few guest lectures in the course.
Registration for this course is by permission only (please contact Prof. Anderson if you're interested). Enrolled students will be required to do their fair share of the class presentations. Additionally, each student team must submit a paper at the end of the course that summarizes the material that they covered. Students will be graded based upon their presentation(s), course paper, and class participation (which will have a greater emphasis in this class than perhaps is usually required). We will have a more thorough discussion of grading in class.
In terms of the department's course classification, this is is an applications course.
The background information classes will be like mini-classes on these topics. They need to emphasize material of importance to automotive use cases, but the point here is to provide needed background, not necessarily to get into automotive issues deeply.
- Getting organized. (Jan. 7, 2015)
- Overview of automotive trends. (Jan. 14, 2015)
Background information, Part I: Real-Time Systems. (Jan. 21, 2015)
- A Very Short Introduction to Real-Time Systems & Real-Time Scheduling Theory by Rui Liu
- Locking Protocol & Multiprocessor Scheduling by Xinghau Xu
- Complexities Arising in Real Systems by Micaiah Chisholm
Background information, Part II: Computer Vision. (Jan. 28, 2015, Feb. 4, 2015)
Presenters: Eunbyung, Yang, Kecheng
Slides blessed by Alex
Background information, Part III: Fault Tolerance and Security. (Feb. 11, 2015, Feb.18, 2015)
Presenters: Namhoon, James, Xinghao (jr), Amos Slides blessed by Jim
Background information, Part IV: Robotics. (Feb. 4, 2015, Feb. 11, 2015)
Presenters: Jeff, Alan, Hannah, Andrew (jr), Dhruv
Slides blessed by Alex
- Introduction to Robotics by Jeffrey Ichnowski
- Motion Planning by Alan Kuntz
- Introduction to Control Theory for Robotics (Keynote) by Dhruv Mittal
- Robotics: From Theory to Physical Systems: Putting it Together by Hannah Kerner
- Spring Break (Mar. 11)
Micheal + Planning (Mar. 18)
(Jim & Sanjoy out of town)
Shige (GM) (Mar. 25)
- Develop Vehicle Control Systems as CPS by Shige Wang (GM R&D)
Student Lead Talk #1: Computer Vision for Automotive CPS (Apr. 1)
- Introduction to Computer Vision Benchmark for Autonomous Driving by Cheng-Yang Fu
- Deep Learning and Computation by Eunbyung Park
- Performance of Computer Vision by Amos Wang
- Student Lead Talk #2: Certification of Automotive CPS (Apr. 8)
Student Lead Talk #3: Real-Time Systems for Automotive CPS (Apr. 15)
(Jim & Sanjoy out of town)
- Scheduling on CAN by Rui Lui
- Response Time Analysis in Real-Time Distributed Automotive Systems by Keycheng Yang
- Real-Time Support for Automotive Application by Xinghao Xu
- Uniprocessor EDF Scheduling for AVR Task Systems by Zhishan Guo
Student Lead Talk #4: Robotics for Automotive CPS (Apr. 22)
Jeff, Alan, Michael, Andrew, Hannah
- Vehicle Localization by Hannah Kerner
- Autonomous Motion Planning for an Automotive System by Alan Kuntz
- Real-Time Motion Planning in Autonomous Vehicles by Jeffrey Ichnowski
- Creating Robotic Platforms by Michael North
James Martin, Namhoon Kim, Dhruv Mittal, and Micaiah Chisholm, " Certification for Autonomous Vehicles". PDF .
Cheng-Yang Fu, Chun-Kun Wang, and Eunbyung Park, " A Survey of Computer Vision Research for Automotive Systems". PDF .
Hannah Kerner, Alan Kuntz, Jeffrey Ichnowski, and Michael North, " Robotics and Autonomous Driving". PDF .
Zhishan Guo, Rui Liu, Xinghao Xu, and Kecheng Yang, " A Survey of Real-Time Automotive Systems". PDF .