AWR6843AOPEVM: Inquiry about AWR6843AOP-EVM suitability for static 3D environment mapping

우현 전

Part Number: AWR6843AOPEVM
Other Parts Discussed in Thread: AWR6843AOP, IWR6843AOP, AWR2944, AWRL6844

Dear TI Support Team,

Hello, our team is currently using the AWR6843AOP-EVM to collect 3D position data of surrounding static objects (e.g., obstacles, walls, ceiling, floor). However, we noticed that most of the SDK example demos are optimized for dynamic object detection and tracking.

We attached our short video showing a person walking. The green points represent detected human movements, which appear to be captured faintly. However, the red points are mostly classified as noise. Below are the parameters we used for the experiment.

6843AOP_Standard2.cfg

We would like to ask a few questions:

Is this radar sensor (AWR6843AOP-EVM) suitable for capturing static environment structures in 3D, or should we consider using a different sensor? we used radar in 2~ 10 meters
Since radar point clouds are relatively sparse, it is often difficult to determine whether a point is reliable or just noise. Would it be feasible to fuse radar data with a depth camera to verify and filter out unreliable points?
Are there any recommended methods or algorithms to distinguish noise from valid static points when using the AWR6843AOP-EVM?

Thank you very much for your time and support. We look forward to your advice.

15 days ago

0 우현 전 15 days ago

Prodigy 20 points

hmm.. i can't upload my video and image. i try to upload but "file embedding is not allowed.. my image is 302*387 jpg

0 Kristien Everett 15 days ago in reply to 우현 전

TI__Genius 14436 points

Hello,

Thank you for reaching out. Please see my responses below to each of your questions.

우현 전 said:
Is this radar sensor (AWR6843AOP-EVM) suitable for capturing static environment structures in 3D, or should we consider using a different sensor? we used radar in 2~ 10 meters

Yes, the AWR6843AOP is capable of detecting static structures, but as you have noticed, most applications are for dynamic object detection and tracking as this is one of radar's advantages. You can take a look at one of our industrial applications that's designed specifically for static object detection, the area scanner demo. This demo was made for IWR6843AOP, but these parts are functionally the same, thus the project can be recompiled for AWR6843AOP if needed.

우현 전 said:
Since radar point clouds are relatively sparse, it is often difficult to determine whether a point is reliable or just noise. Would it be feasible to fuse radar data with a depth camera to verify and filter out unreliable points?

Yes, it is possible as we have looked at camera radar fusion - see the following video.
https://www.ti.com/video/6364706537112

우현 전 said:
Are there any recommended methods or algorithms to distinguish noise from valid static points when using the AWR6843AOP-EVM?

As mentioned previously, I would look at the area scanner demo code for more details on implementation, but at a high-level, a theoretical approach would be to utilize a peak detection algorithm, such as CFAR, within the range dimension of the heatmap and only with the zeroth doppler bin - i.e., non-moving points - to filter only the strongest signals.

우현 전 said:
hmm.. i can't upload my video and image. i try to upload but "file embedding is not allowed.. my image is 302*387 jpg

Videos are not allowed on E2E, but you should be able to upload any images you have.

Hopefully this helps, but let us know if you have any questions.

Regards,

Kristien

0 우현 전 11 days ago in reply to Kristien Everett

Prodigy 20 points

I want to make the sensor detect static points such as walls or the floor when I move it by hand. The conditions are not exactly the same, but I’m trying to test it based on a similar case I found in a forum reply.

https://e2e.ti.com/support/sensors-group/sensors/f/sensors-forum/1534754/iwr6843aopevm-cannot-detect-trihedral-corner-reflector-stationary-object?tisearch=e2e-sitesearch&keymatch=Cannot detect trihedral corner reflector#

However, I keep running into build errors when building the DSS project. The error message indicates that there isn’t enough memory allocated for the fastCode section. Based on LLM’s suggestion, I modified the linker configuration line as follows: load = L3SRAM PAGE 1, run = L2SRAM PAGE 0

But it still seems that the memory isn’t sufficient and the build fails.
Is there any recommended way to resolve this issue?

0 Kristien Everett 10 days ago in reply to 우현 전

TI__Genius 14436 points

Hello,

What SDK version are you using when rebuilding the project? Have you modified it in any way? The project is currently supported by SDK 3.5.0.2, so other SDK versions may cause build issues.

Regards,

Kristien

0 우현 전 8 days ago in reply to Kristien Everett

Prodigy 20 points

I’m currently working with SDK 3.6.2 and CCS 20.3.1, which may be contributing to the issue I’m seeing.

Separately, the TI AWR6843AOP-EVM may not be the best fit for our project, since its software pipeline seems optimized for detecting dynamic objects rather than static structures. Therefore, I’m considering switching to a different radar sensor.

Within TI’s radar lineup, could you recommend a sensor suited for SLAM mapping in indoor environments (walls, floor, ceiling) at short range (≤10 m), where high azimuth resolution is preferred?

0 Kristien Everett 8 days ago in reply to 우현 전

TI__Genius 14436 points

Hello,

우현 전 said:
Within TI’s radar lineup, could you recommend a sensor suited for SLAM mapping in indoor environments (walls, floor, ceiling) at short range (≤10 m), where high azimuth resolution is preferred?

The only other devices that may satisfy your application would be AWR2944 and AWRL6844 as these both utilize a 4Rx 4Tx antenna array, but these devices and examples are built with automotive applications in mind which are typically longer range and motion based. For example, you can watch and read about Provozio's collaboration with TI on the AWR2944 to achieve environmental map: https://www.ti.com/video/6366781952112

Just to clarify, is your end application for automotive or industrial, such as robotics?

Regards,

Kristien

0 우현 전 8 days ago in reply to Kristien Everett

Prodigy 20 points

Hello,

We are planning to use radar to assist indoor human vision .

Is the collaboration version by Provizio Ltd with Texas Instruments on the “AWR2944” a different product from the standard AWR2944?

0 우현 전 8 days ago in reply to Kristien Everett

Prodigy 20 points

++

The specs of the AWR2944 from the Provizio collaboration look outstanding.
However, since Provizio is a solution-providing company, we probably can’t purchase just a single board, right?

0 Kristien Everett 7 days ago in reply to 우현 전

TI__Genius 14436 points

Hello,

Thank you for clarification on the end application. The Provizio board is their own solution, but it utilizes TI's AWR2944 chip for radar sensing. I will loop in one of my colleagues who can comment further.

Regards,

Kristien

Sensors

Sensors forum

AWR6843AOPEVM: Inquiry about AWR6843AOP-EVM suitability for static 3D environment mapping