Single Chip MM-Wave Radar

By: Damien Clarke
Lead Consultant
25th April 2019
4 minute read
Recent advances in radar technology have led to the production of a range of inexpensive highly integrated single chip millimetre-wave radar sensors by Texas Instruments. These chips implement a Frequency Modulated Continuous-Wave (FMCW) radar operating at either 76 – 81 GHz or 60 – 64 GHz. This provides sufficient bandwidth to produce a range resolution of a few centimetres at the same time as measuring object velocities via Doppler shift. In addition, through the use of multiple transmitters and receivers Multiple-Input and Multiple-Output (MIMO) techniques can be used to measure the angular position of an object. With a suitable 2D antenna array it is possible to simultaneously measure both azimuth and elevation angles.
The processing power necessary to calculate the range, velocity and angles to multiple targets is also present within the chips. In the IWR6843, for example, this is achieved via a C674x DSP, an FFT hardware accelerator and an ARM R4F Microcontroller. This also enables the ability to perform object tracking within the chip. A single inexpensive chip can therefore continuously output a point cloud (object ID, range, azimuth, elevation and radial velocity) of multiple unique objects present in the scene.
A common application for such radar sensors is the detection of moving vehicles at a distance. The video below shows an example of two cars driving towards and then away from a radar placed above the road. Raw data is extracted from the chip and processed to emulate what would normally occur within the chip. The left hand chart shows a Range-Doppler map where the two vehicles are clearly detected at all ranges. All static objects have been removed from this image to more clearly reveal moving objects. The central plot shows those range-Doppler cells which are determined to contain non-background objects (i.e. cars). The right hand plot then calculates the 2D position (in metres) of the two cars. Note that a car will produce multiple radar echoes (i.e. radiator, wing mirror, tires, number plate, etc) at different ranges and therefore a cluster of detected points are produced from each car.
The ability of such a sensor to directly output a processed point cloud enables a wide range of possible applications at a low cost. These include the following:
-
- Advanced driver-assistance systems (ADAS)
-
- Autonomous ground vehicles
-
- Unmanned Air Vehicles (UAV)
-
- Traffic monitoring
-
- Pedestrian and people counting
-
- Intruder detection
-
- Vital signs detection
-
- Gesture recognition
-
- Fluid level sensing
Creating a new product using a Texas Instruments mm-wave radar chip requires development in several areas. Firstly, as with all FMCW radar it is necessary to understand what radar configuration to use to achieve the desired data output parameters, i.e. range resolution, max range, velocity resolution, etc. It is also necessary to modify the processing chain implemented on the chip to optimise the performance. Hardware changes will also be required, in particular the design and manufacture of a suitable mm-wave antenna array is of key importance. This will have several effects, but commonly this is used to increase the maximum detection range. Finally, it is also necessary to produce an electronics design for the additional components which must be integrated with the radar chip to create a final product.
Recent advances in radar technology have led to the production of a range of inexpensive highly integrated single chip millimetre-wave radar sensors by Texas Instruments. These chips implement a Frequency Modulated Continuous-Wave (FMCW) radar operating at either 76 – 81 GHz or 60 – 64 GHz. This provides sufficient bandwidth to produce a range resolution of a few centimetres at the same time as measuring object velocities via Doppler shift. In addition, through the use of multiple transmitters and receivers Multiple-Input and Multiple-Output (MIMO) techniques can be used to measure the angular position of an object. With a suitable 2D antenna array it is possible to simultaneously measure both azimuth and elevation angles.
The processing power necessary to calculate the range, velocity and angles to multiple targets is also present within the chips. In the IWR6843, for example, this is achieved via a C674x DSP, an FFT hardware accelerator and an ARM R4F Microcontroller. This also enables the ability to perform object tracking within the chip. A single inexpensive chip can therefore continuously output a point cloud (object ID, range, azimuth, elevation and radial velocity) of multiple unique objects present in the scene.
A common application for such radar sensors is the detection of moving vehicles at a distance. The video below shows an example of two cars driving towards and then away from a radar placed above the road. Raw data is extracted from the chip and processed to emulate what would normally occur within the chip. The left hand chart shows a Range-Doppler map where the two vehicles are clearly detected at all ranges. All static objects have been removed from this image to more clearly reveal moving objects. The central plot shows those range-Doppler cells which are determined to contain non-background objects (i.e. cars). The right hand plot then calculates the 2D position (in metres) of the two cars. Note that a car will produce multiple radar echoes (i.e. radiator, wing mirror, tires, number plate, etc) at different ranges and therefore a cluster of detected points are produced from each car.
The ability of such a sensor to directly output a processed point cloud enables a wide range of possible applications at a low cost. These include the following:
-
- Advanced driver-assistance systems (ADAS)
-
- Autonomous ground vehicles
-
- Unmanned Air Vehicles (UAV)
-
- Traffic monitoring
-
- Pedestrian and people counting
-
- Intruder detection
-
- Vital signs detection
-
- Gesture recognition
-
- Fluid level sensing
Creating a new product using a Texas Instruments mm-wave radar chip requires development in several areas. Firstly, as with all FMCW radar it is necessary to understand what radar configuration to use to achieve the desired data output parameters, i.e. range resolution, max range, velocity resolution, etc. It is also necessary to modify the processing chain implemented on the chip to optimise the performance. Hardware changes will also be required, in particular the design and manufacture of a suitable mm-wave antenna array is of key importance. This will have several effects, but commonly this is used to increase the maximum detection range. Finally, it is also necessary to produce an electronics design for the additional components which must be integrated with the radar chip to create a final product.