Safe Autonomous Cyber Physical Systems Workshop 2016

Safe Autonomous Cyber Physical Systems Workshop 2016


Recent advances in cyber physical systems, such as drones and autonomous cars, combined with the internet-of-things (IoT), have led to a lot of consumer interest. However, to fully unleash the capabilities of such systems we need to invent a new level of safety. For example, an autonomous robot needs to model the environment and take only those actions that will ensure the safety of itself and everything around it, while still making progress on the assigned task.

Addressing this problem entails an interdisciplinary approach, where innovative ideas from perception, sensing and modeling need to work with advances in controls, planning and decision-making. This will be a large-scale effort in software and hardware engineering, and we all know that managing software complexity in large code bases is difficult, so it is important to also improve our programming languages so we can achieve deeper levels of verification and achieve a stronger guarantee of the safety of those systems.

This workshop brought together experts in the areas of machine learning, vision, planning, programming languages and verification to discuss the latest advances with the aim of catalyzing a joint effort between academia and industry with the goal of advancing Safety in Autonomous Cyber Physical Systems. The daylong workshop consisted of invited talks, discussions, and demos as part of the Microsoft Faculty Summit 2016.

Attendee Information

  • Transportation: We ran a shuttle from the Hilton Bellevue and Marriott Bellevue to the MSCC from 7:15 AM–8:30 AM (every 15 minutes) the morning of July 15. Buses also made trips from MSCC to the two hotels every 15 minutes starting at 2:30 PM with the last bus leaving at 5:00 PM.
  • Food & Beverages: We served breakfast, lunch and beverage breaks/snacks. There was a dinner the night before (Thursday, July 14) to celebrate the 25th year anniversary of Microsoft Research. Details of this dinner were provided as part of the welcome packet at the hotel as well as on the faculty summit website.
  • Check-in: Registration at Microsoft Conference Center to obtain stick-on-badge
  • Contacts: Ashish Kapoor (Chair), Debadeepta Dey, and Shital Shah

Speaker Information

  • Check-in: At speaker desk near main entrance. Speakers were requested to sign a release form for the recording of the event and to provide their slides.
  • Gear Note: If speakers had an Apple/Mac laptop, we requested they bring a HDMI adapter and Apple-to-DVI adapter and power cord
  • Speaking time: 30 mins including questions



  Time Session Speaker
8:45 – 9:00
Opening Remarks
Ashish Kapoor
9:00 – 9:30
Driver Adaptation to Increasingly Autonomous Systems
Linda Ng Boyle
9:30 – 10:00
Dependable Computer Vision for Autonomous Systems
Martial Hebert
10:00 – 10:30
Intelligent Vehicles that (Fore) See
Dariu Gavrila
10:30 – 11:00
Break (Poster session in parallel)
11:00 – 11:30
Real-time On-board Localization and Planning for Small UAVs
Roland Siegwart
11:30 – 12:00
Representations vs. Algorithms: Higher Level Autonomy in Populated Environments
Nicholas Roy
12:00 – 12:30
Fast Safe Perception Planning and Control
Debadeepta Dey
12:30 – 1:30
Lunch (Poster session in parallel)
1:30 – 2:00
Safe Learning in Robotics
Claire Tomlin
2:00 – 2:30 Quantitative Policies over Streaming Data Rajeev Alur
2:30 – 3:00
Models and Verification Tools for Cyber-Physical Systems
Rupak Majumdar
3:00 – 3:30
Break (Poster session in parallel)
3:30 – 4:00
Robot Influence on Human Actions
Anca Dragan
4:00 – 4:30
Closing Plenary: Reflections on Safety in AI Systems
Eric Horvitz
4:30 – 5:00
Closing Remarks and Discussion



Title Author Association
Planning for Autonomous Cars that Leverages Effects on Human Actions
Dorsa Sadigh et al.
University of California, Berkeley
Compositional Reactive Synthesis for Multi-Agent Systems Solar Moarref University of Pennsylvania
Decoupling Light Reflex from Pupillary Dilation, to Measure Emotional Arousal in Videos Pallavi Raiturkar University of Florida
Probabilistic Shared Control for Maximizing Automation Success and Safety Parkar Owan University of Washington
Building a system for interacting with swarms of robots Ivan Gavran Max-Planck Institute for Software Systems
Exploring Motion Features for Off-Road Pedestrian Detection Benzun Babu Worcester Polytechnic Institute
Pre-Capture Privacy Sensors for Computer Vision Francesco Pittaluga University of Florida
PLATO: Policy Learning using Adaptive Trajectory Optimization Tianhao Zhang University of California, Berkeley
Learning to Gather Information Sanjiban Choudhury et al. Microsoft Research
Re-planning without Regret: Optimism-Driven Re-planning in Uncertain Environments Wen Sun et al. Microsoft Research
Actor Modules for Safe Asynchrony Ankush Desai et al. Microsoft Research
Continuous-time Safety Certificates under Uncertainty Brian Axelrod et al. Microsoft Research