As ADAS and autonomous driving systems evolve, engineering teams face a convergence of challenges: increasingly complex perception pipelines, surging compute demands, new regulatory expectations and a global market where the speed of innovation varies dramatically from region to region.
Key Topics on the Agenda
AutoSens USA brings together thought leaders and industry experts from OEMs, Tier 1s and suppliers to examine how the next generation of vehicle intelligence will be built and deployed
at scale.
Below, take a closer look at the key themes driving the 2026 agenda.
Perception Robustness Under Real-World Conditions
As automated driving systems move closer to large-scale deployment, the real test of perception is no longer performance in ideal conditions, but resilience in the messy reality of everyday driving. Inconsistent road markings, ageing signage, stray light, glare, adverse weather, and infrastructure designed for human vision all place pressure on perception stacks in ways that are difficult to replicate in controlled environments.
This track brings together insights from autonomous truck operators already deploying systems at scale, who will share the real-world challenges they encounter and the methods they use to maintain reliable perception in demanding conditions. Their perspectives are complemented by input from key stakeholders including IEEE working groups, academic researchers, and meteorologists, offering a multidisciplinary view on how environmental factors, standards, and sensing assumptions intersect. Together, these contributions highlight how robust perception is being engineered, validated, and evolved beyond the lab, informed by lived operational experience and cross-domain expertise.
Architectures & Edge Compute for Safety-Critical Vehicle Intelligence
Edge compute is becoming the decisive frontier for automotive AI. As perception networks grow and sensor bandwidth increases, the strain on vehicle compute platforms is intensifying. OEMs face competing pressures: enable more intelligence at the edge, reduce system latency, comply with safety standards and keep BOM cost under control.
This track examines the architectural decisions shaping the next wave of domain controllers and distributed compute systems. Expect discussions on hardware acceleration, memory bandwidth constraints, deterministic execution, ASIL-compliant processing paths and real-time orchestration of heterogeneous workloads. The industry is grappling with how to maintain safety guarantees while deploying increasingly complex AI models within tight thermal, power and latency budgets. The goal: compute architectures that scale without collapsing under their own weight.
Edge-Case Validation & Assurance at Scale
Autonomy fails not in the common scenarios but in the rare ones, the edge cases that are unlikely, unpredictable, or poorly represented in datasets. As perception stacks become more AI-driven, traditional validation strategies struggle to keep up.
This track focuses on how engineers uncover unknown unknowns, measure safety across vast scenario spaces and combine simulation with proving ground and on-road testing. Discussions will explore world modelling, agent-based scenario search, dataset curation, synthetic edge-case generation and the emerging regulatory expectations for demonstrating safety. As systems grow more complex, the challenge is no longer just detecting failures, it’s proving that you know where your system might fail and that you have predictable, safe fallback behaviours when it does.
Emerging Modalities
At the sensor level, innovation hasn’t slowed. High-resolution radar is reaching near-LiDAR fidelity in some use cases, hyperspectral sensing is opening new possibilities for material classification and novel photonics and frequency-modulated approaches are extending perception into domains once thought impractical.
This track explores which of these technologies will reach production scale, what gaps they fill in the perception ecosystem and how they integrate into existing stacks. With adverse weather and low-visibility conditions still posing major performance constraints, emerging sensors may determine which OEMs can reliably deliver next-generation ADAS features across global markets, not just in ideal conditions.
The Minimalist Vehicle – Lowering Latency, Compute, and Cost
As OEMs look to scale ADAS across more models and markets, efficiency becomes key. This track examines how to design perception systems that use leaner models and more efficient compute pipelines, delivering high performance without prohibitive cost or power draw. The challenge is to identify what can be removed without degrading reliability and where the trade-offs lie between redundancy, system robustness, and affordability.
The pressure to democratise ADAS globally is rising and OEMs must deliver performance that is both economically and energetically sustainable.
From Software-Defined to AI-Defined: The Next-Gen Vehicle Intelligence Stack
A generational shift is underway. At AutoSens China 2025 we heard from one keynote speaker
“The SDV is already dead, the future is AI Defined” – but can the West keep up with China’s technological pace? Here, instead of static, rules-based stacks, vehicles are moving toward
AI-defined architectures built on generative models, agentic systems, world understanding
and adaptive behaviours. This track looks at the emerging approaches reshaping how vehicles learn, plan and improve, marking a transition from software-defined to AI-defined intelligence to remain competitive in the global market.
AutoSens USA 2026 will showcase the people and technologies building the foundations of safer, smarter and more efficient autonomy. These tracks reflect not just where the industry is today, but where it’s going.
DID YOU KNOW?
AutoSens USA ticket holders for will also gain access to the InCabin USA conference. InCabin is the leading event for the Intelligent Cockpit, covering driver and occupant monitoring, active and passive safety systems, HMI
and UX, and many more.
