Software BARK

Software BARK

Tutorial Benchmarking Interactive Behavior Planning for Self-Driving Vehicles using BARK

The tutorial, which will be held during the International Conference on Intelligent Robots and Systems (IROS) in Las Vegas, Nevada (USA) on 25 October 2020, which this year will be virtual due to the current corona situation. The event is for anyone interested in becoming familiar with innovative benchmarking approaches for use with interactive decision-making algorithms.

BARK is an open source benchmarking platform designed for facilitating the comparison of approaches developed in the field of interactive behavior planning for autonomous vehicles. The tutorial offers an opportunity to more closely observe and discuss this unique approach, in which behavior models can be used interchangeably for forecasting, planning and simulation purposes. The event also aims to promote the application of the BARK platform in research and industry projects.

fortiss scientists Julian Bernhard, Klemens Esterle, Patrick Hart and Tobias Kessler will provide an insight into the following issues:

  • Benchmarking behaviors
  • Autonomous driving
  • Multi-agent systems
  • Interactive planning

The event offers an excellent starting point for discerning the status of research activities in this field and provides an understanding of the current challenges and opportunities for future research strategies.


October 25, 2020
9:00 am to 1:00 pm









Due to the current Corona situation, the event will take place virtually, participation is free of charge.

Target groups

Doctoral students (robotics, software engineering), experienced robotics researchers in science and industry, industry specialists


Required knowledge

Required knowledge Participants should have at least basic knowledge of decision-making processes - meaning the underlying principles of decision problems such as MDPs and POMDPs - understand and have experience with Python and have a general understanding of C++.



  • Principles of state-of-the-art approaches for interactive behavior planning (POMDP planning, multi-agent planning)
  • Pros and cons of modern approaches
  • Understanding the requirements of benchmarking platforms in order to systematically develop interactive behavior planners
  • Hands-on experience using BARK as a benchmarking platform for interactive planners
  • Implementation and benchmarking of the participants’ own interactive behavior planner

The exact schedule of the tutorial, plus a list of presenters and additional information is available here.

Registration online

Follow the link below to register for the tutorial “Benchmarking interactive planning for self-driving vehicles using BARK”, as well as for the 2020 IROS conference.

Click here to register

 Julian Bernhard

Your contact

Julian Bernhard

+49 89 3603522 583