Hello, I’ve recently purchased the mmWave Human Detection kit, at first everything worked fine, but after trying out the extended features (motion speed, distance sensing, etc.) everything started breaking.
The “Body Movement Parameter” started spiking at random intervals (to levels anywhere between 40 and full 100) and would not drop to 0 even if no one was in the room and the sensor was facing the external wall of the building, which in turn meant that “Presence Information” always showed “Detected”.
That’s unworkable levels of dirty data.
The only sources of interference in my room I could think of were my PC, a few power supplies, my Bluetooth headset and cables lying around.
In addition, ESPHome does not recognize that the device is online, only when on a wired connection, and when looking at logs it shows that it has successfully connected to my Wi-Fi network and has an IP address, but when trying to access logs wirelessly I get the following message: "INFO Starting log output from seeedstudio-mmwave-kit.local using esphome API WARNING Can’t connect to ESPHome API for seeedstudio-mmwave-kit.local: Error resolving IP address: [Errno -5] No address associated with hostname (APIConnectionError)"
I have tried reinstalling the factory firmware from the website and flashing a new .yaml configuration to it (copied directly from the github “mr24hpc1.yaml”).
I’ve been troubleshooting the device for the past 2.5 hours and I’m tired and completely out of ideas.
Could someone help me out with this?
If not I’ll have to return it, which is sad, because its features are really great and I was looking forward to using it in automations.
P.S.: the parameter spiking was also observed while looking at the logs, turning off/on advanced settings did nothing to help
Hi, thanks for the response.
Honestly that connection issue is the least of my concerns, the fact it gets recognized won’t do much if the sensor itself is malfunctioning, but I’ll give it a shot.
Any pointers for that DNS resolver? Could it be a problem form the HA side or the Wi-Fi AP? And can I just give ESPHome the IP address of the device instead of trying to resolve it via DNS?
Unfortunately I can confirm parts of the concerns. I just bought the kit some days ago and no matter what parameters I try I get random spikes of the body movement and it never drops below 1 - with nobody around.
I tested so many parameter combinations that I am pretty sure it is an issue with the firmware.
I have managed once to get the movement to 0 and thus the presence to false by setting the detection range to a ridiculous low level of 1m. But that of course defeats the purpose of a presence sensor.
So it’s not just a me problem, good.
Now the question is should we await a firmware update/try and figure out a fix ourselves (if feasible) or should I just return the sensor and look for something less exciting, but more stable.
I’ve also sent this issue to their tech support email, I’ll wait for their answer and keep you posted.
I am out of ideas to be honest. I have tried the build in scenes. I have used the custom mode. Different places in the room. When that thing is directly facing a wall and still reports presence… well.
As long as nobody from seeed considers this to be an issue a firmware update will not appear.
I personally have ordered a different device already as I am not patient enough with so many other products on the market that work.
But I am keeping the seeed one around in case some miracle happens
Well now that I think about it, technically I think it should be possible to just erase the Body Movement Parameter and general presence detection out of the configuration and determine presence off of other values, that are reported correctly.
Still, one of the major features of the presence sensor being almost completely nonfunctional is a big no-no in my books.
But hey, you gotta make due with what you have.
I’d have to think that over, but maybe a combination of static and motion distance, filtered existence and motion energy, although I’d have to monitor them if they don’t spike too, but so far only the Body Movement Parameter was all over the place, the rest seemed ok on the first glance.
I just got a Tuya ZY-M100 and after tuning the parameters for about 10 mins I got the desired result.
So I am really not sure if that Seeed kit is worth the trouble.
Probably not, but since it costed me a pretty penny I’m experiencing a strong case of sunk cost fallacy.
I’ll see how big of a window do I have to return it and I’ll try my luck untill then, if I won’t make it work by then, then it’s getting returned.
If you have already confirmed the correct installation method, please ensure that the measurement environment is a single-person indoor space. Please check whether the Active reporting of presence information is constantly reporting the data below regardless of the presence of a person: Report: 0x53 0x59 0x80 0x01 0x00 0x01 0x01 sum 0x54 0x43
If it is, please try adjusting the sensitivity.
I made modifications to the Motion Trigger Boundary inquiry parameter, adjusting it from 2m to 5m. I used the upper computer software to monitor the motion parameters, (Motion Energy waveform). The testing was conducted in a spacious indoor environment. Before the modification, the waveform was close to 0 beyond a distance of 2m. After the modification, the waveform also approached zero beyond a distance of 5m.
If possible, we suggest conducting tests in a spacious indoor environment with only one person. If the issue still persists, please feel free to contact [email protected] for further assistance.
Thank you for the response, but unfortunately, no amount of adjusting the parameters changes the behavior I’ve described.
The Body Movement Parameter is simply unusable in its current state.
The problem is not with other measurements the sensor is performing, but just with the Body Motion Parameter, which you claim uses a special algorithm, my guess is that this algorithm needs some serious adjustments.
I admit that my use case is not the ideal scenario for the sensor, my room is rather short (2m at the tallest point) and more elongated than square, but still, stuff like that should not happen with a product marketed as relatively “plug and play”.
I’d like to return the kit, but troubleshooting it took me so long that the 5 day return window since receiving it has passed, is there a way for me to return it anyway? I don’t want to just throw it away, that would be a waste.
I’ve had the same issues. I’ve been testing this kit three days now and the only thing that’s consistent is how unreliable “Presence Information” is. Regardless of settings and what it measures, it gets stuck on either “Detected” or “Clear,” and the only thing that makes it change is a system reset. I’ll just try a different sensor, this one doesn’t work.
Very sorry for the inconvenience!
If you want to perfactly realize the functions of the kit, you need to use the mmWave kit according to the above installation method and usage environment. If the data fluctuates a lot, it is recommended that you can remove the radar, connect it to your computer with a usb to serial port, and adjust its sensitivity or other parameters according to the datasheet to adjust your radar so that it can be suitable for your own usage scenario.
I have exactly the same problem here, Body Movement Parameters jumping up and down randomly. Maybe a hint to the seeed engineers - when removing the cover the kit works perfectly. Maybe some kind of parasitic self oscillation …? interesting that this issue wasnt posted before january 2024
We’re very sorry guys, we’ve noticed that people are giving us feedback on this issue, and the R&D department is already collecting materials to check the cause of the problem. Please give us a few days and when I have new information, I will sync it to you as soon as possible.
PJ
