Inside a laboratory in Warren, Michigan, General Motors is watching you sweat. Not metaphorically — literally. As part of an ambitious research programme aimed at delivering fully hands-off, eyes-off autonomous driving by 2028, GM is strapping sensors to test subjects and using artificial intelligence to measure fear, confusion and stress in ways that subjects themselves cannot articulate.
The goal is the next evolution of Super Cruise, GM’s existing hands-free highway driving system, which already allows drivers to remove their hands from the wheel but still requires them to keep their eyes on the road. The successor system, slated to debut on the electric Cadillac Escalade IQ, will go a step further: letting drivers divert their attention from the road entirely under certain conditions.
Inside the Simulator
The research takes place in a purpose-built simulator equipped with seven projectors casting imagery across a curved screen that wraps around a physical replica of a Cadillac Lyriq cabin — complete with a real steering wheel, pedals and seatbelts. Roads, skies, buildings and roadside signage are all digitally generated, creating a controlled environment in which engineers can introduce any scenario they choose: sudden obstacles, extreme weather conditions, erratic behaviour from other road users.
Test subjects are fitted with sensors on their head and hands, along with a pulse oximeter on one finger. The system records eye movements and pupil dilation, heart rate and heart rate variability, and galvanic skin response — the change in electrical conductance caused by stress-related perspiration. Steering angle, brake pressure and indicator use are also logged.
According to GM, the facility runs millions of simulations each day, the equivalent of tens of thousands of days of human driving.
The Lie Detector for Drivers
The central question GM is trying to answer is deceptively simple: how long does it take a driver to regain control when an autonomous system hands back responsibility?
Omer Tsimhoni, GM Technical Fellow for display and optics research, has noted that the mental transition involved in moving from an eyes-off state back to active driving is not easily captured through post-test questionnaires. Drivers may report feeling fine while their biometrics tell a different story. If a subject ticks the “no issues” box but their heart rate and skin conductance spiked at the moment of handover, GM’s team takes the objective data as the more reliable account.
The company refers to this approach as Emotional AI — algorithms that interpret human emotional states from facial expressions, vocal tone and physiological signals. During eyes-off test sessions, subjects are given a tablet to play a game, simulating the kind of real-world distraction a driver might engage in if their attention were no longer required. The critical moment comes when the system signals that it needs the driver back: GM is studying which combination of auditory, visual and haptic alerts produces the fastest and least stressful handover.
The Road to 2028
GM’s R&D lead on this effort, Linda Cadwell Stancin, comes from an aerospace background, with more than two decades split between Boeing and Lockheed Martin. Her experience with aviation autopilot systems — which manage comparable human-to-machine handovers at altitude — is understood to be shaping GM’s methodological approach to the problem on the road.
The Cadillac Escalade IQ, already positioned as GM’s flagship electric SUV, is the intended launch vehicle for the eyes-off system when it arrives in 2028.
Beyond Autonomy
The Warren campus is also host to parallel research threads that point toward GM’s broader ambitions. Engineers are developing lithium manganese-rich batteries — lower in energy density than conventional NMC cells, but cheaper and less reliant on scarce minerals like cobalt. A virtual wind tunnel powered by AI is being used to deliver instant aerodynamic drag predictions, replacing computational fluid dynamics simulations that traditionally take weeks to run. Alok Warey, Lab Group Manager for R&D, has indicated that aerodynamic drag can account for up to half of an electric vehicle’s energy consumption — making the tool strategically significant for range optimisation.
Atomistic simulations are also being used to model real-time changes in battery chemistry and understand their effects on performance.
GM has been headquartered on the Warren campus since 1956 and accounts for roughly four percent of Michigan’s GDP. In 2026 the company extended its footprint into Formula 1 with the Cadillac entry into the world championship — another arena where marginal gains in engineering and human-machine interaction carry considerable stakes.
