Elon Musk has acknowledged that Tesla’s Robotaxi service is struggling to operate late into the night because its vehicles have trouble spotting small animals in the dark — a challenge highlighting long-standing criticisms of the company’s camera-only autonomous driving system. Addressing why the service has been slow to expand after dark, Tesla CEO pointed to low-light, low-contrast detection as the key obstacle, specifically citing the difficulty of identifying “grey kittens on grey tarmac.”The remarks came after Tesla announced recently that its Austin Robotaxi pilot would extend operating hours by another one hour, moving its night time up from 10:00 PM to 11:00 PM. Despite the incremental gain, the service remains more constrained than when it debuted in June 2025 with operating hours spanning 6:00 AM to midnight. Musk quoted a post from Robotaxi handle on X, formerly Twitter that announced Robotaxi being now available till 11 pm, Musk wrote, “Robotaxi operating hours moved from 10pm to 11pm. The main thing we’re trying to solve is making sure that we don’t run over pets when they’re hard to see at night. Literally trying to avoid grey kittens on grey tarmac in the dark.”
Elon Musk on Tesla’s ‘kitten problem’
The technical bottleneck Elon Musk described exposes a classic limitation of optical cameras. As passive sensors, cameras depend entirely on ambient illumination and contrast to distinguish objects from their background. A low-contrast subject, such as a grey animal against dark asphalt, reflects minimal light and provides little distinct visual data.By contrast, active sensors like lidar and radar do not rely on ambient lighting. Lidar projects its own laser pulses to generate precise, real-time 3D depth maps, detecting any object that protrudes above the road surface regardless of colour or darkness. Radar emits radio waves to track distance and velocity, cutting through fog, dust, and glare that routinely blind camera lenses.Commercial autonomous fleets operating at scale — like Google-parent Alphabet’s Waymo — rely on a multi-sensor suite combining cameras, lidar, and radar to run 24-hour operations across thousands of vehicles. Also in contrast to Robotaxis, Waymo’s driverless vehicles run 24 hours a day.
How Google’s Waymo has solved ‘kitten problem’
In August this year, Waymo co-CEO Dmitri Dolgov demonstrated this capability by releasing footage of children and dogs running across an unlit street; while the optical cameras struggled with visibility, the lidar feed tracked the subjects clearly. Dolgov made the comments during a talk at Y Combinator’s Startup School as he walked the audience through the lessons Waymo has learned building its driver over close to two decades.Tesla, however, has pursued a strictly vision-only strategy under Musk’s direction. Musk has spent years dismissing competing hardware setups, famously calling lidar a “fool’s errand” in 2019 and later claiming that combining sensors causes conflicting data that reduces overall safety. Tesla stripped radar units from its production vehicles in 2021, phased out ultrasonic sensors in 2022, and has consistently avoided lidar, even saying that anyone relying on it is “doomed.“Incidentally, safety regulators are scrutinising the company’s approach. The US National Highway Traffic Safety Administration (NHTSA) is actively investigating 3.2 million Tesla vehicles over FSD collisions in low-visibility environments. As part of that inquiry, federal investigators requested internal company communications, including a document titled “Radar Saves Us”, recalling Musk’s private admission in 2021 that vision paired with high-resolution radar would outperform vision alone.