IN Brief:
- Gryphon uses 48 transmitting and 48 receiving channels to produce 2,304 virtual radar channels.
- The published sensor specifications include a 300m detection range, 20 frames per second and 25W consumption.
- Arbe positions the product for military and civilian ground autonomy without announcing a Gryphon defence order.
Arbe Robotics has introduced Gryphon, an ultra-high-resolution 4D imaging radar designed to help uncrewed ground vehicles interpret rough, unstructured terrain. The system is designed to map obstacles and traversable space when dust or poor light restrict optical sensors and when satellite navigation signals are degraded or unavailable. Arbe identifies military and civilian applications, including logistics and industrial vehicles, but has not disclosed a military purchase order, fielding programme or completed integration into a named ground platform for Gryphon.
The sensor uses 48 transmit and 48 receive channels to create 2,304 virtual channels. Multiple combinations of transmitting and receiving antennas provide angular information that can be used to distinguish nearby returns from different objects. The radar produces a spatial point cloud with associated velocity information, rather than reporting only the presence of a moving target. That detail is required by slow-moving ground vehicles negotiating stationary walls, vegetation, debris and road edges as well as moving traffic.
Arbe lists an angular resolution of 0.7 degrees in azimuth and 0.9 degrees in elevation, with a field of view extending 120 degrees horizontally and 30 degrees vertically. Its published detection range is up to 300 metres, and the sensor generates as many as 20 frames per second. These manufacturer-stated figures do not establish the detection probability of a small hazard at 300 metres, performance amid severe clutter or the safety of a vehicle’s subsequent navigation decisions.
Gryphon is described as supporting free-space mapping, odometry and localisation as well as obstacle detection. Free-space mapping attempts to identify traversable parts of the scene, whereas odometry estimates movement over time from sensor observations. Localisation then uses those observations and an available map or other references to estimate the vehicle’s position. Where global navigation satellite system reception is poor, the radar estimates remain susceptible to accumulated error, changes in the scene and ambiguous reflections, requiring calibration and validation against other information.
The practical challenge is not limited to detecting large vehicles. A small autonomous carrier may need to identify a roadside ditch, a hanging obstruction or rough ground in which wheel traction could be lost. Arbe says Gryphon’s dense point cloud can support identification of negative obstacles such as holes and washouts, alongside rocks, poles and vegetation. Detection alone cannot prove a route safe, because ground strength, vehicle clearance, wheel placement and trajectory planning still determine whether the vehicle can pass.
Cameras and lidar can provide high spatial detail but may lose useful returns when heavy dust or fog blocks the optical path. The radio-frequency measurement can preserve environmental information under conditions that scatter or obscure visible and laser light, subject to the radar’s own operating limits. It has its own limitations, including multipath reflections, material-dependent responses and difficulties separating weak returns from background clutter. A system built around Gryphon would still need carefully tested perception software and, depending on the mission, other sensors to establish reliable operation outside laboratory conditions.
Arbe publishes a minimum operating range of 0.5 metres, 10–80 centimetre range resolution depending on configuration, and velocity resolution of 0.1 metres per second. The company specifies a 25-watt power requirement, an approximately one-kilogram unit and dimensions of 135 by 103 by 46 millimetres. The radar is listed with an IP67 enclosure rating and an operating temperature range from minus 40 to plus 70 degrees Celsius. Together, the electrical load, physical dimensions, enclosure protection and temperature rating constrain vehicle mounting, power supply and heat-management choices.
The interface is specified as 1000BASE-T Ethernet, providing a familiar route for moving radar data into a vehicle computer. Ethernet connectivity does not ensure that existing vehicle autonomy software can interpret the radar’s point-cloud format without interface development and calibration. Time synchronisation, sensor orientation, vibration isolation and the location of the antenna relative to other equipment all affect the resulting picture of the surroundings. Vehicle designers would also need to assess electromagnetic compatibility and reliability under the platform’s actual mechanical loads.
Gryphon extends Arbe’s radar activity into off-road autonomy following earlier company agreements involving other radar applications. In July, Arbe announced a collaboration with an unnamed defence and homeland-security integrator covering three projects, and subsequent company reporting referred to deliveries against that relationship. The earlier integrator announcement concerned existing Arbe radar technology and did not identify Gryphon as a contracted product. Arbe’s separately introduced Alerion counter-drone radar addresses aerial threat detection rather than the ground-vehicle perception role specified for Gryphon.
Ground autonomy needs a combined approach to sensing, trajectory selection, controls and safe fallback behaviour. Higher angular resolution can provide more spatial detail than conventional obstacle-warning radar, but the autonomy controller still needs a tested method for deciding when observations are reliable enough to act on. Arbe’s announcement supplies a detailed engineering specification and identifies potential integrators; it does not supply independently measured field performance, an accepted military qualification or production volumes. Those remain the evidence needed to assess adoption beyond the product launch.
