Test, Measurement & Inspection Product of the Year
Product Name: Radar Scene Emulator
Company: Keysight Technologies
While the possibilities of autonomous vehicles are enticing, developing complex systems that outperform even the best human drivers is an immediate challenge that automakers must solve in order to realize this vision of mobility in the future.
Existing solutions in the marketplace all had two things in common. First, they used a product called a radar target simulator (RTS) to emulate distance. These bench-top instruments are available in various forms, including from Keysight. Second, they used mechanical motion to emulate the changing positions of the objects. Some used giant robot arms each holding a large RTS, while others were slightly more elegant, moving just antennas coupled through cables to a few stationary RTSs. These brute-force solutions worked, but because they were mechanical, they were clumsy, and the number of targets was limited, forcing the reduction of a complex scene into just a handful of simplified points. Imagine trying to orchestrate tens or hundreds of targets moving independently of each other using robotic arms! There had to be a better way.
The Keysight team set out to invent a whole new class of emulators, something capable of realistically reproducing a scene with no moving parts. The team evaluated multiple approaches before settling on one using a 2-D array of radar pixels, which we now call rixels. Much like a 2-D array of pixels on a monitor can emulate real-time motion in fine detail, the 2-D array of rixels would do the same. The rixels had to be close together to achieve the resolution required for realistic simulation, yet the solution needed to emulate many spatially distinct targets over a wide field of view.
Currently available radar target simulators were benchtop-sized instruments. Automakers certainly couldn’t cram hundreds of these together into a lab. The Keysight team leveraged our unique capabilities in monolithic microwave integrated circuits (MMIC) design to replicate the functions of a benchtop instrument in a small thumbnail-sized microcircuit. This innovation made it possible to have hundreds of miniaturized radar target simulators to create the Radar Scene Emulator.
Keysight’s first-to-market technology combines hundreds of miniature radar target simulators into a scalable screen that can emulate objects with up to 512-pixel resolution and at distances as close as 1.5 meters. This breakthrough radar scene emulation solution overcomes conventional radar sensor test solutions that have a limited field-of-view (FOV) and cannot simulate objects at distances less than 4 meters.
Utilizing “total scene generation,” the radar scene emulation solution exercises your automated drive systems and algorithms by applying time-synchronized inputs to the actual sensors. Its open architecture also closes the loop with your existing hardware-in-the-loop (HIL) systems and 3D modelers. These capabilities create a solution that complements—and fills the gap between—software simulation and on-road testing. As such, it overcomes limitations of software simulation that does not test real radar sensor response, while achieving repeatable testing of radar scenes, which cannot be done on the test track.
The Radar Scene Emulator allows you to emulate real-world driving scenarios, varying speed, distance, and number of targets, across a contiguous FOV. With radar sensors and back-end software confidently tested against the complexity of real-world driving scenarios, you’ll achieve your vision of ADAS and next-generation vehicle autonomy sooner, with less risk.
Keysight’s Radar Scene Emulator offers in-lab full radar scene emulation:
1) Thoroughly exercise radar sensors and systems with up to 512-pixel resolution a contiguous horizontal FOV of ±70 degrees and vertical ±15 degrees
2) Supports short, medium, and long-range mmWave radars by generating static and dynamic targets at ranges of 1.5 meters to 300 meters and with velocities of 0 to 400 km/h
3) Address multi-target, multi-angle scenarios with mechanically fixed RF Front ends that provide repeatable angle of arrival (AoA) accuracy
Emulate complex, RF-dense urban scenes with realistic interference testing
4) Using 3D point clouds and multiple reflections allows for improved detection and differentiation of objects
The Radar Scene Emulator provides a deterministic environment for in-lab testing of complex scenes that today can be tested only on the road
Keysight is dedicated to making the world a better place, by pledging to achieve net zero emissions in corporate operations by the end of fiscal year 2040. Keysight’s commitment to attaining this aim of improving the global environment is exemplified via the Radar Scene Emulator, which enables automakers to reduce road testing and pollution by thousands of kilometers every year.