Taking the "what-ifs" out of self-driving cars

Audi Autonomous Driving Cup 2018

Cooperative autonomous vehicles management with learning-based algorithm in VANET

TU Ilmenau will present future technologies on the Hannover Messe

   

We present autonomous driving with high-tech model cars (Audi Q2 on a scale of 1:8). The Audi Autonomous Driving Cup is an international competition aiming at developing fully automatic driving functions and corresponding software architecture, for which our Team AFILSOP won the 1st prize in 2017. Decided for our win is the application of a model-based optimal control alogrithm. Since our algorithm considers the passenger´s desire, road situation and car dynamics, the computed driving strategy is more reliable and comfortable in comparison to the conventional methods. The realized concept is industrial-oriented and can be easily transferred to real cars.

press release

Video Interview

Audi Autonomous Driving Cup 2017

Theory: 3B methods for autonomous driving

   

Below is the Theoretical Approach.

Our practice sets are (click for details):

Goal: realization of autonomous driving

 
   

B1: Data-based (to identify current driving conditions, permissible range, etc.)

  • Laser scanner, radar
  • Camera
  • GPS
  • Communication between vehicles
 

B2: model-based (for predicting the driving dynamics in the next time moment)

  • Model equations
  • Calibration of the parameters
  • Description of the restriction of the permissible range
 

B3: algorithm-based (to calculate the driving strategy in real time)

  • Solution of an optimization problem
  • Computational efficiency
  • Convergence properties

Formulation of the optimization problem

 

Literature

   

Evgeniya Drozdova, Siegbert Hopfgarten, Evgeny Lazutkin, Pu Li
Autonomous Driving of a Mobile Robot Using a Combined Multiple-Shooting and Collocation Method
IAV 2016 - 9th IFAC Symposium on Intelligent Autonomous Vehicles, 29.06.2016 - 01.07.2016, Leipzig, Germany