2-day session - Sensing and Perception
Location
The 2-day session will take place in Linköping. The room Ada Lovelace (former called Visionen) is in the B-building.
Schedule
Feb 22
10.00-10.30 Coffee outside Room "Ada Lovelace" (former Visionen)
10.30-11.00 Welcome and information
11.00-12.00 Presentation: G. Hendeby - "Making Sense of Observations: examples of sensor fusion Download Making Sense of Observations: examples of sensor fusion"
12.00-13.15 Lunch at Universitetsklubben
13.15-14.00 Presentation: Scania (Christoffer Norén, Laura Dal Col and Bengt Boberg)- "Scania’s long term vision for autonomous vehicles, the current focus and the challenges in sensing and perception Download Scania’s long term vision for autonomous vehicles, the current focus and the challenges in sensing and perception".
14.15-15.00 Presentation: I. Skog - "Positioning techniques -- Transforming sensor data to reliable position information Download Positioning techniques -- Transforming sensor data to reliable position information."
15.00-15.30 Coffee outside Ada Lovelace (Visionen)
15.30-16.00 Introduction to entrepreneurial challenge Links to an external site. (BoB)
16.00-19.00 Entrepreneurial challenge (groups and instructions) Links to an external site.
19.00- Dinner, pitches and announcement of winner at Universitetsklubben
Feb 23 Location Ada Lovelace (Visionen)
08.15-10.00 Deep Visual Object Tracking Download Deep Visual Object Tracking, M. Felsberg
10.00-10.30 Coffee outside Ada Lovelace
10.30-11.30 Presentation: K. Åström - "Robust estimation using radio, audio and video Download Robust estimation using radio, audio and video"
11:30-12.15 Presentation: Autoliv - "Autoliv Vision System – One Step Closer to Autonomous Driving"
12.15-13.15 Lunch at Universitetsklubben
13.15-14.00 Presentations from batch1 students
- Per Boström-Rost
- Martin Lindfors
14.00-15.00 Information about the examination step of Module 1 Links to an external site., Q&A Links to an external site. and outlook of Module 2 Links to an external site..
Abstracts
G. Hendeby - "Making Sense of Observations: examples of sensor fusion"
Accurate information about position, orientation, and other properties of a system are often considered known when developing higher level applications. However, in many cases, for different reasons, these properties cannot be directly measured well enough. The solution is to instead estimate the properties from indirect measurements using sensor fusion. In this talk the principles behind this will be outlined, and different techniques with be illustrated using examples from our research.
I. Skog - "Positioning techniques -- Transforming sensor data to reliable position information."
The lecture presents an overview of basic positioning sensors and positioning techniques. First the four key performance indicators – accuracy, integrity, availability, and continuity of service – of a positioning system are introduced. Then the basic properties of exteroceptive and proprioceptive sensors, and how these properties relates to the key performance indicators, are explained. This is followed by a description of feature based positioning, geometric based positioning, and dead-reckoning positioning, which are the most commonly used techniques to transform sensor measurements to position information. Finally, different techniques for fusing the information from multiple positioning sensors and systems are discussed.
Autoliv - "Autoliv Vision System – One Step Closer to Autonomous Driving"
Autonomous driving is currently a very hot field in research. Vehicle manufacturers are also putting a considerable amount of resources into development in this area. Cameras have a central role in these kinds of systems and the requirements are very high on detecting and tracking objects, such as vehicles. This presentation will give some information about vision systems. Focus will be on object detection and especially the target tracking problem, where detections in the image should be used for tracking and estimating object pose and velocity in world coordinates.