Arbe introduces Gryphon radar for autonomous ground vehicles

Arbe Robotics has introduced Gryphon, a four-dimensional imaging radar system developed for autonomous ground vehicles operating on rough terrain and in conditions that can impair optical sensing. The radar uses 48 transmitting and 48 receiving channels to generate measurements of surrounding objects, with applications spanning unmanned ground vehicles, industrial equipment and autonomous off-road platforms.

Announced on 19 October, Gryphon extends Arbe’s radar technology into ground autonomy applications requiring obstacle detection, navigation and vehicle localisation. The company identifies defence, mining, construction, agriculture and logistics as potential markets, particularly where vehicles operate outside the structured environments associated with conventional road transport.

The radar architecture combines 48 transmitters and 48 receivers, providing 2,304 virtual channels through the processing of signals associated with different transmitting and receiving antenna combinations. The resulting measurements include range, azimuth, elevation and radial velocity, which together form the four-dimensional data used to describe the surrounding environment.

Arbe specifies a detection range up to 300m and a minimum range of 0.5m. The stated field of view covers 120 degrees horizontally and 30 degrees vertically, with angular resolution of 0.7 degrees in azimuth and 0.9 degrees in elevation. The system also provides 0.1m/s velocity resolution and an output rate of 20 frames per second.

These are published device characteristics rather than guarantees that every object will be detected at the maximum range. Radar performance depends on target size, material, orientation, distance, surrounding clutter and the installation of the sensor. The ability to distinguish objects also depends on the signal processing and the way radar information is interpreted by the vehicle’s control system.

Radar determines properties of its surroundings by transmitting radio-frequency signals and measuring reflections. The relationship between transmitted and received signals provides information about distance, while observations across antenna elements allow the direction of a return to be estimated. Doppler measurements provide information about relative radial velocity.

With multiple transmitting and receiving channels, the system can process combinations of observations to form a larger virtual antenna arrangement. This provides angular information without requiring every virtual channel to correspond to a separate physical antenna element. The result can support a denser representation of the environment than simpler radar systems designed principally to detect the presence or distance of a small number of targets.

Off-road navigation places different demands on sensing from ordinary highway driving. A vehicle may have to distinguish rocks, excavations, vegetation, terrain changes, personnel and stationary machinery without clearly marked lanes or predictable road boundaries. Some obstacles remain stationary relative to the surroundings, limiting the extent to which Doppler information alone can separate them from background returns.

Gryphon is intended to preserve spatial information that can assist with these decisions. Arbe lists object information, four-dimensional data and high-frame-rate point clouds among the system’s outputs, alongside information intended for free-space estimation and odometry. The precise value of each output depends on the algorithms used to convert radar measurements into a representation suitable for vehicle control.

Free-space estimation concerns identifying regions that may be available for the vehicle to traverse. A radar system can provide measurements of objects and terrain features, but deciding whether an area is safe requires further interpretation. Vehicle dimensions, ground clearance, wheel or track configuration and the stability of the terrain all influence whether an apparent route is usable.

Odometry provides estimates of vehicle movement over time. Radar observations can contribute information about the relative motion of objects and the sensor, supplying another input to localisation algorithms. The resulting position estimate may also incorporate information from inertial sensors, wheel measurements, satellite navigation or other systems.

Where satellite signals are obstructed or unavailable, such sensor fusion can provide alternative information about movement and position. However, changes in terrain, sensor calibration and accumulated measurement errors affect the accuracy of the estimate. Gryphon’s provision of odometry-related data does not establish that the radar can independently maintain precise vehicle position under every operating condition.

Dust is one of the environments specifically addressed by Arbe’s design. Cameras and LiDAR systems rely on optical transmission and reception, which can be affected by airborne particles, obscured surfaces and lighting conditions. Radio-frequency sensing operates at longer wavelengths and can retain useful measurements in some conditions where the performance of optical equipment is reduced.

Radar also has limitations. Strong reflections, multipath propagation and complex terrain can produce ambiguous measurements, while small or poorly reflecting objects may be difficult to detect consistently. The system must distinguish relevant obstacles from environmental clutter, and different operating conditions may require adjustments to the processing used to interpret the returned signals.

Gryphon is a solid-state radar without mechanically rotating scanning assemblies. Its published dimensions are 135 × 103 × 46mm, with a stated weight of approximately 1kg and power consumption of 25W. Arbe lists operation from -40°C to +70°C, IP67 ingress protection and shock and vibration characteristics associated with ISO 16750-3.

Those figures help establish the physical requirements for integration into a vehicle. A 25W sensor still requires appropriate power distribution and thermal management, while its mounting arrangement must maintain a suitable sensing field without exposing connectors or cables to excessive mechanical stress.

IP67 describes defined protection against dust ingress and temporary immersion under specified test conditions. It does not establish resistance to every form of pressure washing, chemical exposure or mechanical damage. Similarly, a component-level environmental specification does not independently qualify a complete vehicle installation.

Data transmission uses a 1000BASE-T Ethernet interface, allowing the radar to connect to an external computing platform. The volume of information transferred depends on the selected output configuration and processing arrangements. High-density data can create additional demands on the vehicle’s network, storage and perception software.

Several sensors may be required to provide coverage around an autonomous vehicle. Combining their outputs involves calibration, synchronisation and the management of overlapping measurements. Additional processing is also needed to prevent separate radar units from producing inconsistent environmental representations when the vehicle changes position.

Gryphon is distinct from Arbe’s Alerion counter-UAS radar introduced in September. Alerion targets the detection and tracking of airborne threats, while Gryphon is intended for the sensing requirements of vehicles moving across difficult ground. The products draw on related radar technology but address different operational problems.

Arbe has not identified a named production customer, disclosed Gryphon shipment volumes or published independent validation of complete autonomous driving performance using the new radar. Integration trials will determine how its sensing characteristics contribute to navigation and obstacle avoidance across the environments identified by the manufacturer.