On Nov 5, 2026, Sony (China) Ltd. globally debuted two mobility technologies at the ninth China International Import Expo: "in-car passenger motion sickness detection fusing dual-channel visual information" and "vision-language-model-based driving takeover risk alert." The two technologies target in-car entertainment comfort and intelligent-driving human-machine interaction respectively. Both involve AI models and visual analysis, and the related technologies have been patented or entered real-road testing.
Motion sickness detection: dual-channel vision, reusing ADAS cameras
According to IT Home, Sony's passenger motion sickness detection system analyzes two video streams at once: one from the road ahead, reflecting acceleration, deceleration, turning and bumping; the other from content playing on the rear-seat screen, identifying visual factors that may trigger motion sickness such as overall motion, rotation and flicker. An AI model trained on public medical data combines both streams to predict the passenger's likely level of motion sickness, and can also assess it against individual sensitivity.
The system reuses the vehicle's existing ADAS forward camera and requires no additional rear-seat camera. The related technology has been patented. Sony China Research Institute developed the system and worked with Beijing Jiaotong University to further address ride comfort in in-car entertainment. The assessment results can prompt the driver to reduce hard braking and sharp turns, and can also give content platforms a reference for recommending content with gentler visual changes.
From driver to passenger, Sony's exploration of mobility has extended to the actual experience of in-car entertainment.
As background, at the eighth CIIE Sony showcased an in-car entertainment experience integrating its "Inspiration Window," in-car audio and PS5 games. Responding to visitors' feedback about motion sickness when watching screens on the move, Sony launched the detection system this year.
Takeover alert: VLM judges ODD boundaries, issues tiered prompts
The other global debut, "vision-language-model-based driving takeover risk alert," uses a vision language model (VLM) to analyze the road ahead and judge whether conditions are approaching or beyond the intelligent driving system's operational design domain (ODD), giving the driver tiered prompts. The takeover prompts have three levels: when approaching the boundary, the driver is reminded to prepare to take over; when beyond the range but not yet urgent, the driver is guided to take over proactively; in high-risk situations such as sudden obstacles or perception failure, an emergency takeover is prompted.
During development, the team used a digital twin simulation environment to generate data for daytime, nighttime, heavy rain and high-speed driving conditions for model training and validation. The technology is now in real-road testing with Tsinghua University's School of Vehicle and Mobility in Xiangcheng District, Suzhou, and can support automakers and intelligent-driving solution providers in judging takeover timing and designing human-machine interaction.
What to watch
Both Sony technologies are at the debut or testing stage, with no mass-production timeline announced. Key follow-ups include the patent layout and automaker cooperation progress of the motion sickness detection system, the road-test data performance of the takeover alert system in Suzhou, and whether mainstream automakers or intelligent-driving solution providers will adopt the two technologies.