Vehicle-Human Interaction Lab

Human-Centered Computing @ Graz University of Technology

Mission & Vision

At the Vehicle-Human Interaction Lab, we pioneer human-centered AI to advance high-performance tasks by fostering a bidirectional exchange between human expertise and artificial intelligence. Our mission is to understand the cognitive and physiological processes underlying elite human performance and leverage this knowledge to design AI systems that learn from and with humans. By integrating physiological sensing, cognitive modeling, and deep reinforcement learning, we aim to create AI agents capable of adaptive decision-making and skill transfer in dynamic, high-paced environments. Our work enhances safety, sustainability, and performance by testing AI-driven innovations in high-fidelity simulations and real-world applications, ensuring that AI not only learns from the best human strategies but also helps refine and extend human capabilities.

We envision a future where humans and AI collaborate seamlessly, pushing the boundaries of high-performance tasks. Through interdisciplinary research, we seek to develop AI that augments human abilities, fosters safer and more efficient decision-making, and accelerates sustainable innovation in autonomous systems and beyond.

Key Research Areas

Human Performance Modeling

Investigating the physiological and cognitive processes underlying elite human performance, leveraging sensor data and computational models to enhance decision-making, adaptability, and human-AI collaboration in high-paced environments.

AI Drivers for Autonomous Racing

Developing high-performance AI drivers through deep reinforcement learning, testing innovations in high-fidelity simulations, and facilitating bidirectional skill transfer between humans and AI for safer, more efficient, and sustainable mobility.

Latest Achievements

The Racer's Gaze Keypoints

We introduce a novel concept of Gaze Keypoint events integrating eye, head, telemetry, and localization data.

Learn More

Autonomous Racing Benchmark

Our team created a high-fidelity benchmark with AI algorithms for autonomous racing.

Learn More

Human Telemetry

Our team developed a platform for physiological data collection synchronized with high fidelity simulation.

Learn More

Publications

Publication Thumbnail

Uncertainty-Based Experience Replay for Task-Agnostic Continual Reinforcement Learning

Remonda, Terrel, Veas, Masana, Transactions on Machine Learning Research. IN PRINT

Publication Thumbnail

The Racer's Gaze: Visual Strategy in High-Speed Sports Expertise

Lappi, Pekkanen, Krajnc, Iacono, Remonda, Veas, UNDER REVIEW

Publication Thumbnail

A Simulation Benchmark for Autonomous Racing with Large-Scale Human Data

Remonda, Hansen, Raji, Musiu, Bertogna, Veas, Wang, Neural Information Processing Systems NeurIPS, 2024

Publication Thumbnail

Multimodal decoding of error processing in a virtual reality flight simulation

Wimmer, Weidinger, Veas, Müller-Putz. Sci Rep 14, 9221 (2024)

Publication Thumbnail

Comparing driving behavior of humans and autonomous driving in a professional racing simulator

Remonda, Luzhnica, Veas. (2021) PLOS ONE 16(2): e0245320

Publication Thumbnail

Formula RL: Deep reinforcement learning for autonomous racing using telemetry data

Remonda, Krebs, Luzhnica, Veas. (2019) IJCAI Workshop on Scaling-Up Reinforcement Learning: SURL, 2019

Collaborators and Groups Working in this Lab

The Vehicle-Human Interaction Lab is utilized by various research groups and teams working on cutting-edge projects:

Interested in our research or want to collaborate? Reach out to us and join our mission to redefine elite performance.

Get in Touch