• State Resolved
  • Locked Locked
  • Replies 19 replies
  • Answers 2 answers
  • Subscribers 32 subscribers
  • Views 487 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

Original question:

MMWCAS-DSP-EVM: Point cloud sparse problem

MMWCAS-DSP-EVM: MMWCAS-DSP-EVM object detection issue

Chennan Xu
Prodigy 95 points
Part Number: MMWCAS-DSP-EVM
Other Parts Discussed in Thread: AWR1843, MMWCAS-RF-EVM, TIDEP-01012, AWR2243

Hi,

I am using MMWCAS-DSP-EVM to detect 2 metallic water bottle (high reflectivity), and these 2 water bottle are only 130cm away from the radar sensor. However, from the mmWaveStudio post processing plots by the scripts "cascade_MIMO_signalProcessing.m" from the "4chip_cascade_MIMO_example" folder, and also the matlab script I wrote, I think they are not being detected. I tried to zoom in the range-velocity plots and range-azimuth plots, but there is nothing detected in the range of 130cm.

  • 0
    Prodigy 95 points

    I accidentally inserted 3 repeated pictures, just ignore two of them.

  • 0 in reply to Chennan Xu
    TI__Mastermind 24755 points

    Hi,

    We will need to check on this with the team internally. We will get back to you on this by early next week.

    Thanks,

    Pradipta.

  • 0 in reply to Pradipta Bandyopadhyay
    TI__Mastermind 19515 points

    Hi Chennan,

    The demo with its configuration was intended to detect objects such as cars, two wheelers, pedestrians etc. The reflected signals from the water bottle were weak therefore it ended up being merged into the noise floor from the looks of it. 

    • Can you isolate your test environment such that the bottle's signals do not suffer from the near far problem?
    • Would it be possible for you to use a corner reflector instead for evaluation?

    Regards,

    Kaushik

  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    Hi Kaushik,

    I tried to use another single-chip radar AWR1843 to detect the same water bottles that I use in this scenario, but it works and shows high reflectivity. I do not have a corner reflector at the moment but I will try to isolate the test environment to see whether it works.

    Regards,

    Chennan

  • 0 in reply to Chennan Xu
    TI__Mastermind 19515 points

    Hi Chennan,

    Please do and let me know how it goes.

    .

    Regards,

    Kaushik

  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    Hello Kaushik,

    I tried to detect a object in isolated test environment for 4 different cases, which are 1 meter away from the radar sensor, 2 meters away, 3 meters away and 3.5 meters away from the radar sensor with 15 degrees between them. The radar cannot detect the object for the first two cases but it can detect the object precisely for the last two cases. 

    May I assume that the MMWCAS-RF-EVM is not very robust for near-field objects detection? But why this is the case? I think the radar do not have minimum detection range.

  • 0 in reply to Chennan Xu
    Prodigy 95 points

    I tried to collect a few more sets of data, it seems like all the results are quite similar and it does not make any sense at all. Everytime before I start capturing, I create a new folder by changing the capture_directory   =   "" in Lua script to store the data, so it is unlikely to use the same data for post-processing.

    Detailed procedure of my data capture process:

    1. run the Lua scripts Cascade_Configuration_MIMO.Lua first to load the radar configurations.

    2. run the Lua scripts Cascade_Capture.Lua to capture data.

    3. Check the directory /mnt/ssd in WinSCP to see if there is a new folder created with the same name as in step 5, and also make sure the data is stored in this new folder.

    4. Transfer the file to my laptop for post-processing.

    5. Change the variable "capture_directory" in Lua scripts Cascade_Capture.Lua and move the target objects to change the test environment.

    I repeat the above procesure for 3 times, the pictures of test environment are attached below.

    a. test environment 1 (2 objects are placed around 3 meters away from the sensor)

    b.test environment 2 (2 objects are placed very close to each other)

    c. test environment 3 (one object is placed 3 meters away and the other one is placed around 5 meters away)

  • 0 in reply to Chennan Xu
    TI__Mastermind 19515 points

    Hi Chennan,

    Can you let me know what your range max and range resolution definitions are for your chirp design?

    .

    Regards,

    Kaushik

  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    Hi Kaushik,

    According to the document "Design Guide: TIDEP-01012 Imaging Radar Using Cascaded mmWave Sensor Reference Design", the max range for the MIMO configuration is 150m.

    The range resolution can be calculated by using the equation c/(2B) which is equal to 0.0475. 

    Also, the chirp configuration defined in Cascade_Configuration_MIMO.lua is shown below, I change the number of frames to 3, and number of chirps to 64, and remain the rest of parameters as default.

    local profile_indx              =   0
    local start_freq                =   77                              -- GHz
    local slope                     =   79                              -- MHz/us
    local idle_time                 =   5                               -- us
    local adc_start_time            =   6                               -- us
    local adc_samples               =   256                             -- Number of samples per chirp
    local sample_freq               =   8000                            -- ksps
    local ramp_end_time             =   40                              -- us
    local rx_gain                   =   48                              -- dB
    local tx0OutPowerBackoffCode    =   0
    local tx1OutPowerBackoffCode    =   0
    local tx2OutPowerBackoffCode    =   0
    local tx0PhaseShifter           =   0
    local tx1PhaseShifter           =   0
    local tx2PhaseShifter           =   0
    local txStartTimeUSec           =   0
    local hpfCornerFreq1            =   0                               -- 0: 175KHz, 1: 235KHz, 2: 350KHz, 3: 700KHz
    local hpfCornerFreq2            =   0                               -- 0: 350KHz, 1: 700KHz, 2: 1.4MHz, 3: 2.8MHz

    -- Frame configuration  
    local start_chirp_tx            =   0
    local end_chirp_tx              =   11
    local nchirp_loops              =   64                              -- Number of chirps per frame
    local nframes_master            =   3                               -- Number of Frames for Master
    local nframes_slave             =   3                               -- Number of Frames for Slaves
    local Inter_Frame_Interval      =   100                             -- ms
    local trigger_delay             =   0                               -- us
    local nDummy_chirp              =   0      
    local trig_list                 =   {1,2,2,2}                       -- 1: Software trigger, 2: Hardware trigger

    Regards,

    Chennan

  • 0 in reply to Chennan Xu
    Prodigy 95 points

    Hi Kaushik,

    In Lua script "Cascade_capture", I noticed that I set the "n_files_allocation = 0;" , which might cause frame drops while switching files. I am not sure if this is the cause of the problem.

    Also, the explanation of this parameter written in Lua script is, 
    -- n_files_allocation: is the number of files to preallocate on the SSD.
    --      This improves capture reliability by not having frame drops while switching files.
    --      The tradeoff is that each file is a fixed 2047 MB even if a smaller number of frames are captured.
    --      Pre-allocate as many files as needed based on (size_per_frame * number_of_frames) to be captured
     
    Also, can you please explain the last sentence in detail?
    Regards,
    Chennan
  • 0 in reply to Chennan Xu
    TI__Mastermind 19515 points

    Hi Chennan,

    • Can you confirm if you are observing frame drop at the boundaries of this transition into different files?
    • Your resolution estimation seems to be incorrect. For a max range of 150m, the resolution would be much greater than 4.5 cm. Can you please check that? In any case, you will need to make sure your object in the test can be resolved based on this value or reduce the resolution.
    • The last statement means that the number of files to be allocated is to be decided based on the total size of your data that is expected i.e., let's say you are capturing 100 frames of size 1MB each, then you would need 1 file pre-allocated as 1 file can hold up to 2GB of data. 

    .

    Regards,
    Kaushik

  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    Hi Kaushik,

    I think there are two modes for the MMWCAS-RF-EVM (MIMO and TXBF)and I am using the MIMO specification. For the MIMO specification, the max range of 150m is stated in this document "Design Guide: TIDEP-01012 Imaging Radar Using Cascaded mmWave Sensor Reference Design", and its range resolution is 60 cm. However, I do not know how to derive this range resolution, from my perspective, the range resolution is only related to the bandwidth, which is 3159.44MHz in mmWaveStudio. If I want to use MMWCAS-RF-EVM for short-range detection, may I ask what parameters or settings I should configure?

    Regards,

    Chennan

  • 0 in reply to Chennan Xu
    TI__Mastermind 19515 points

    Hi Chennan,

    Would recommend you go through the following document.

    https://www.ti.com/lit/an/swra553a/swra553a.pdf

    Regards,

    Kaushik

  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    Hi Kaushik,

    I have go through the document you provide, the default parameters in Cascade_Configuration_MIMO.lua script is suitable for short range application. The parameters are,

    slope                     =   79 MHz/us
    ramp_end_time             =   40  us
    sample_freq               =   8000 ksps
    so, bandwidth can be calculated as, 
    Bandwidth = slope*ramp_end_time = 3160 MHz
    Therefore, using the following 2 equations,
    The max range is 15.2m and range resolution is 0.0474m.
    Is that correct?
    Regards,
    Chennan
  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    Also, I tried to collect some new data, but the post-processing results are still very similar with the results I obtained a few days ago (The position of the object does not change), although the test environment is completely different.

    Regards,

    Chennan

  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    For example, there are no object in test environment, but there is still object detected for all of the 4 devices in multiple frames. .

    Regards,

    Chennan

  • 0 in reply to Chennan Xu
    TI__Mastermind 19515 points

    Hi Chennan,

    This is strange indeed. I'm not sure if there is anything in your surrounding that corresponds to the detected object at 2.5ish m? The wall perhaps? Is this observation consistent if you completely obstruct the field of view of the radar? Say, by placing a metallic plate right in front of the board.

    Before you experiment with your bottle, I suggest you free up your test environment of all possible clutter.

    Regards,

    Kaushik

  • 0 in reply to Kaushik Gowda
    Prodigy 95 points

    Hi Kaushik,

    Placing a metallic plate right in front of the board, I will try this.

    Also, will it be a problem caused by the software and firmware version? I am using mmwavestudio 2.1.1.0, and mmwave DFP 2.2.4.0 for the four-device cascaded array of AWR2243 devices.

    Regards,

    Chennan

  • 0 in reply to Chennan Xu
    Prodigy 95 points

    The problem is caused by the software version, I should use mmwave studio 3.0.0.14.