MicroVision and FieldAI are extending their collaboration on lidar perception for autonomous machines after tests involving ground vehicles and aerial mapping. FieldAI has purchased additional MOVIA sensors for development and evaluation across its government and industrial programmes, creating a further route to assess MicroVision hardware alongside the autonomy software used on different robotic platforms.
FieldAI’s federal team has evaluated the IRIS long-range lidar sensor on autonomous ground vehicles, while a separate demonstration paired MicroVision’s MOVIA Air Plus with a Robinson Unmanned Spirit aircraft. That aerial system produced real-time, colourised three-dimensional lidar maps displayed through the Android Team Awareness Kit (ATAK). The ground and aerial activities were established before the expanded collaboration; the new development includes the sensor purchase and wider evaluation across FieldAI’s teams.
Lidar uses emitted light and the timing or other characteristics of returned signals to estimate distance to surrounding surfaces. A sequence of measurements can be assembled into a spatial representation of the scene, allowing navigation software to distinguish free space from obstacles and compare the observed environment with a planned route. A vehicle or aircraft needs a suitable combination of sensing range, field of view and update rate for its particular movement pattern.
On ground vehicles, the IRIS evaluation addresses the time available to perceive an obstacle before the machine reaches it. Greater sensing range can provide earlier data for route planning, provided the observations can be interpreted quickly enough for the vehicle’s speed and operating environment. Vegetation, uneven terrain, weather and occlusion all affect the information reaching the controller, which must distinguish useful features from ambiguous returns.
MOVIA sensor evaluation also addresses close-range movements. A robot working among stacked materials, walls, other vehicles and people needs information about nearby obstacles while turning, positioning or approaching equipment. MOVIA sensors are being assessed for applications that can include these conditions. The new purchases support development trials, with integration decisions and customer deployment volumes still to be established.
Aerial mapping uses the same distance-measurement principle from an elevated, moving platform. The resulting point cloud records three-dimensional geometry, and associated image information can add colour to help interpret structures, routes or terrain. In the Robinson demonstration, maps were integrated with ATAK for viewing and situational awareness; that display integration is distinct from the autonomous flight-control software.
Robotic systems operating outside controlled factory cells often encounter irregular ground, changing vegetation, dust, weather and objects that were absent during initial mapping. Their control software therefore needs to deal with uncertain observations as well as clear geometric features. FieldAI describes its autonomy approach as combining data-driven models with physical reasoning and awareness of uncertainty, while MicroVision contributes sensing hardware and perception tools.
FieldAI aims to use a common autonomy approach across ground and aerial platforms, but the sensing arrangement still has to be adapted to each machine. Mounting position, vibration, weight, power demand, processing capacity and communications links affect the data available to the controller. A wheeled industrial transporter negotiating a loading area therefore needs a different installation and validation plan from an aircraft mapping the same site from above.
The partners are assessing long-range, short-range and aerial sensing combinations rather than nominating a single arrangement for every platform. Environmental testing and software integration will establish whether the data are sufficiently consistent for each robot’s navigation functions. A suitable lidar return provides geometry; the system also needs logic for classifying obstacles and choosing a permitted route.
Autonomous machines working close to people and equipment also require safe stopping, access controls and defined operating procedures. The lidar system can supply information used by navigation and obstacle-detection software, while any safety-related functions must be engineered and validated for the complete machine. Those practical constraints will influence how FieldAI uses the purchased sensors during the evaluation phase.
FieldAI can use the additional MOVIA units to gather representative measurements, vary installation positions and test perception in the environments targeted by its industrial and government programmes. Those trials will provide evidence for hardware selection and software configuration before a platform moves towards routine operation.
MicroVision is applying its lidar portfolio to autonomous robotics beyond its established automotive development work, with engineering teams in the United States and Germany. Its current sensors are being tested across industrial, security and government programmes whose navigation and mapping requirements differ from those of passenger vehicles.
The next development milestones are platform-specific integration decisions and repeatable performance during field trials. MicroVision has reported successful evaluations and a sensor purchase, while broader adoption will depend on how reliably the sensor combinations support FieldAI’s autonomous machines under their intended operating conditions.
