Human vs. AI: Pushing the Limits of Autonomous Racing

Nov 16, 2025 · 2 min read
projects

A major milestone in autonomous driving is proving that software can rival human intuition and reaction times at the absolute dynamic limit. This project highlights the significant technical milestones we achieved as a team through years of continuous development.

Key Technical Achievements

While winning championships proves system reliability (see my other post), demonstrating raw pace against a human benchmark makes the technology tangible. Our system achieved several impressive feats:

  • The “Man vs. Machine” Benchmark: Our autonomous race car completed a lap of the Yas Marina Circuit in 58.2 seconds - an autonomous lap record that came within just 0.7 seconds of former Formula 1 driver Daniil Kvyat’s benchmark (57.5 seconds).
  • Absolute Speed: As seen in the Fastest Lap video linked above, the car seamlessly navigates complex F1 corners with centimeter-precision at over 250 km/h.
  • Dynamic Overtaking: Shown in our Sprint Race Overtake video, the stack successfully plans and executes high-speed overtakes - something which has never been done before!
  • Incident-free competitive operation over extended high-intensity race weeks.

A Team Effort

These capabilities were only possible because we worked together for years. Translating cutting-edge research into a robust stack on track was the culmination of relentless effort across perception, planning, control, and systems engineering.

My specific contribution within this collective effort centered around:

  • Designing the software architecture and defining critical system interfaces to handle extreme data loads with low latency.
  • Strategic planning and prioritizing features needed to safely push the dynamic limit in public.
  • Designing and Improving our Control Algorithms to reach almost human-like peformance.
  • Managing testing cycles that allowed us to transition from conservative prototypes to outperforming the competition.

Public Impact

Demonstrating incredible speed and clean overtaking maneuvers in front of a live audience built immense public trust. The performance was featured on the popular German TV show Galileo, bringing the realities of modern autonomous racing to millions through a clear, intuitive “Human vs. AI” comparison.

Simon Sagmeister
Authors
Research Associate and Teamlead
Team Lead and Researcher with 4+ years of experience building real-world autonomous driving systems. Passionate about reliable, high-performance, and modular robotics software, AI, simulation, and motion control.