IN Brief:
- Murata and Smart Eye are combining automotive-grade 60GHz radar with AI-based camera processing for in-cabin sensing.
- Radar adds depth and obscured-occupant detection alongside the positional and behavioural information available from cameras.
- The concept could reduce dependence on separate occupancy sensors while supporting restraint control, child detection, and future software-defined functions.
Murata and Smart Eye are combining automotive-grade 60GHz radar with AI-based camera processing in a concept in-cabin sensing platform intended to improve occupant monitoring, child-presence detection and future software-defined vehicle functions.
The collaboration brings together Murata’s RF and millimetre-wave radar expertise with Smart Eye’s driver and occupant monitoring software. The two technologies provide different views of the same cabin: cameras can interpret posture, head position and visible behaviour, while radar adds ranging information and can continue detecting occupants when lighting is poor or parts of the body are obscured.
That makes the engineering proposition one of sensor fusion rather than simple redundancy. A camera can provide richer visual classification, but its performance is affected by darkness, occlusion and the physical position of the occupant. Radar is less dependent on those conditions, although the data it generates requires signal processing and classification before it becomes useful to the vehicle.
Combining the two sources gives the system a better chance of distinguishing whether a seat is occupied, where a passenger is positioned and whether a child is present even when blankets, clothing or cabin layout make visual detection more difficult.
Murata says the approach could also reduce reliance on dedicated occupancy technologies such as weight mats and capacitive sensors. That would simplify some vehicle architectures and potentially reduce component count, although replacing established occupancy sensors would depend on the safety functions allocated to the fused system and the validation required by each vehicle programme.
Occupant position is becoming more important as restraint systems gain finer control. Airbag and seatbelt deployment can depend on whether a passenger is sitting normally, leaning forward or positioned unusually close to an airbag module, while reliable child-presence detection has become a separate safety requirement in several markets.
Radar is attractive in that environment because 60GHz devices can detect small movements and provide distance information without visible illumination. Compact antennas also allow the RF system to be packaged behind trim or into confined locations where an additional visible sensor would be difficult to accommodate.
Cameras bring the opposite strengths and weaknesses. They offer detailed contextual information, including head direction, body posture and observable behaviour, but generate larger data volumes and are more exposed to lighting and occlusion problems. Fusing radar and vision allows each technology to cover conditions in which the other is weaker.
The trade-off is additional integration work. Radar and camera streams have to be synchronised, calibrated and processed within a common coordinate system, while the vehicle controller needs sufficient compute capacity to interpret both without compromising latency or power consumption.
Packaging also matters. Radar field of view, antenna placement and RF interaction with cabin materials have to be considered alongside camera sight lines and illumination. A concept that performs reliably in one cabin cannot simply be assumed to behave identically in a vehicle with different trim, seating geometry or sensor locations.
Murata and Smart Eye are positioning the platform for software-defined vehicles in which one sensing installation can support several functions over the life of the car. A common sensor set could provide inputs for restraint control, child detection, driver monitoring, gesture recognition and other cabin applications, with software determining how the hardware is used.
That architecture has obvious appeal to vehicle manufacturers, but it makes the boundary between sensor hardware and inference software more important. Updates may change classification behaviour without altering the physical sensors, creating a requirement for controlled software validation alongside the original electronics qualification.
The current system remains a concept rather than a production component. Murata states that specifications and appearance may change, and the companies have not announced a customer programme or series-production date.
The collaboration nevertheless illustrates the direction of cabin electronics. Driver monitoring, occupant detection and restraint sensing are beginning to converge around shared perception hardware, making sensor fusion, compute allocation and software validation central vehicle-design issues rather than separate subsystem decisions.
