What changed
IBM's HRL Laboratories has developed a low-SWaP (size, weight, power) radar system designed to maintain high performance while drastically reducing hardware costs and energy consumption.
The system targets applications such as drones, autonomous vehicles, and distributed sensing networks, where traditional radar systems are often too large, power-hungry, or expensive.
The goal is to bring radar systems into the tens of thousands of dollars range, rather than hundreds or millions, marking a significant cost shift for the industry.
What this means in practice
The low-SWaP radar system rethinks radar as an integrated system, combining semiconductor devices, algorithms, and signal-processing software. This approach allows for more efficient use of commercial off-the-shelf (COTS) components, which are modified to meet specialized radar requirements without losing their economic advantages.
The system also leverages machine learning to filter radar data at the edge, reducing the amount of information that needs to be processed and improving efficiency.
The radar measures objects by transmitting radio-frequency (RF) signals and analyzing the returning pulses for information about distance and motion. Unlike cameras, radar does not produce recognizable images but instead provides data on object location and speed.
Machine learning is used to identify relevant data patterns, with Kim estimating that 80% to 90% of radar data may be irrelevant to higher-level tasks.
Limitations and availability
The supplied announcement does not specify the exact hardware requirements or licensing conditions for the low-SWaP radar system. While the system is designed for applications with limited space, power, and computing resources, the availability of the technology is not detailed.
The focus remains on co-designing radar hardware and software to achieve efficiency without sacrificing performance. The approach emphasizes using efficient algorithms and hardware to meet strict power and computing constraints, rather than optimizing every component to its theoretical limit.
