This workshop will be held at the IEEE Intelligent Vehicles Symposium (IV) 2026, on June 22nd, 2026, in Michigan, USA.

Location: TBD

Scope and Topics

This tutorial provides a practical introduction to implementing robust state estimators for automated driving systems (ADS). On a broad view, the tutorial is divided into two parts: a proprioceptive observer design for ADS, and an efficient dynamic-aware visual odometry algorithm (DynaNav-SVO). The tutorial discusses the mathematical background, optimization tools, robust observer design principles, graph optimization, multimodal data fusion for autonomous navigation subject to disturbances and uncertainties in onboard perception systems. The tutorial merges theoretical and practical aspects of the:

  1. Dynamic vehicle model and tire forces
  2. Sensor fusion of proprioceptive sensory stream and motion models
  3. Combined-slip effect incorporation in visual-inertial navigation in ADS and field robotics
  4. Integrating road surface information in state estimation
  5. Feature-based visual odometry and bundle adjustment
  6. Handling dynamic scenes with exteroceptive sensors

Agenda

Time Talk Title    
04:00:00 Opening ceremony    
04:10:00 Introduction of a novel proprioceptive speed/slip estimation approach by including additional states from the wheel dynamics    
04:30:00 Developing a wheel-vehicle/robot dynamical model for designing an integrated kinematic-tire-based observer framework for robot’s/vehicle’s state estimation    
04:50:00 The design procedure for optimization- and filter-based observers to include the slip states essential for control and safety-critical decision-making in ADS.    
05:10:00 Extensive simulation studies and road experiments will be demonstrated.    
05:50:00 Refreshments break    
06:10:00 Visual odometry: problem formulation and similarities with Simultaneous Localization and Mapping (SLAM)/Structure-from-Motion (SfM)    
06:30:00 Feature extraction and correspondence search    
06:50:00 Motion estimation as an optimization problem (with theoretical backgrounds)    
07:10:00 Static scene segmentation for robust Point-N-Points/Bundle Adjustment    
07:30:00 Integration with proprioceptive state estimator for high-slip scenarios (in field robotics and ADS) and under tire force nonlinearities    
07:50:00 Closing Ceremony    

Organizers

  • Marcelo Contreras, University of Alberta
  • Ehsan Hashemi, University of Alberta

Feel free to contact the organizers if you have any question.