Revision History

Changes from C Revision (November 2019) to D Revision

• Added in 3D spaceGo
• Changed 16m to 15mGo
• Changed greater than 90% to of +- 10cmGo
• Changed 3 people to 1 personGo
• Added ODS + ICB, and AOP (Antenna on Package) Go
• Added 3D People Counting TI Resource Explorer FolderGo
• Changed /ICB Bundle to BoardGo
• Deleted Ideal for Environmental EffectsGo
• Added and Elevation field of View of 44 DegreesGo
• Changed 16m to 15mGo
• Deleted Which can Extend to 14 m with Different Chirp ConfigurationGo
• Changed Azimuth to Approximate azimuthGo
• Changed 44 degrees to 40 degreesGo
• Changed Status Clutter to capon beamforming signal chainGo
• Changed 16 to 15Go
• Changed counting to trackingGo
• Deleted countGo
• Added appliance controlGo
• Changed Radars allow an accurate measurement of distances, relative velocities of people, and other objects to Radar provides accurate position and velocity measurement of people in people in the area of interest.Go
• Added elevation estimation,Go
• Added 4 dimensionalGo
• Added The signal chain follows the mmWave SDK DPM structureGo
• Changed IWR6843 ISK to IWR6843ISKGo
• Added and IWR6843ISK-ODSGo
• Changed ...structure. This information is passed on to the group tracker for object localization. to ...structure all the way through tracker and data output.Go
• Added with the mmWave SDK architecture.Go
• Changed 30° to 40°Go
• Changed 6.3 to 8.1Go
• Changed 5.5 to 8.4Go
• Changed 23.6 to 16.2Go
• Changed 0.3246 to 0.324 Go
• Deleted foundational components that help end users focus on their applications. The SDK also provides several demonstration applications, which serve as a guide for integrating the SDK into end user mmWave applications. This reference design is a separate package, installed on top of the SDK packageGo
• Added a foundational structure for developing radar processing software. This structure includes a Data-Path Manager (DPM) which handles execution of the DataPath - Processing Chain (DPC). The DPC is made of Data Path Units (DPUs). This demo implements custom DPUs within a custom DPC to achieve the people coutning software. Go
• Added Implemented on HWA and Cortex R4EGo
• Changed beam forming to Beamforming (BF)Go
• Changed inverse angle spectrum generation to angle spectrum generationGo
• Added Implemented on c674 DSPGo
• Added Implemented on c674 DSPGo
• Deleted Perform target localization, and report the results.Go
• Added Implemented on Cortex R4EGo
• Deleted The mmWave SDK is divided into two broad components: mmWave Suite and mmWave Demos.Go
• Added The mmWave SDK is the foundational software components of the mmWave projects and includes the following smaller componentsGo
• Added componentsGo
• Added For more information, see the mmWave SDK user's guide.Go
• Deleted mmWave processing demonstration mmWave Demos ADC data capture demonstration and ADC data streaming demonstration from the SDK provides a suite of demonstrations that depict the various control and data processing aspects of an mmWave application. Data visualization of the output of a demonstration on a PC is provided as part of these demonstrationsGo
• Deleted sentence in first bulleted listGo
• Changed 6 m and 14 m to 8mGo
• Changed 60.6 to 60.75Go
• Changed 6.3 to 8.1Go
• Changed 5.5 to 8.4Go
• Changed 23.6 to 16.2Go
• Changed 0.36 to 0.324Go
• Changed 10 to 25Go
• Changed 50 to 55Go
• Changed 2713.66 to 1780Go
• Changed 53 to 57Go
• Changed 2.50 to 2.95Go
• Changed 128 to 96Go
• Changed 128 to 288Go
• Changed 2.355 to 2.55Go
• Changed 24 to 28 Go
• Changed 176 to 239Go
• Changed 640 to 768Go
• Changed 600 to 720Go
• Added Doppler power map to -azimuth heat map.Go
• Changed It is a less time sensitive program. Data can also be stored here to It also stored system code not required to be executed at high speed.Go
• Changed Currently unused to 20Go
• Changed Shared memory buffer between the DSP and the R4F relays visualization data to the R4F for output over the UART in this design to Shared memory bugger used to store slow, non-runtime code.Go
• Changed ...implemented as DSP code executing on the C674x core in the IWR6843.to ...implemented on both the DSP and Cortex R4F. In the following section, we break this process into the following smaller blocks:Go
• Changed Range Azimuthto Range FFT through Range AzimuthGo
• Added image title Range FFT through Range-Azimuth HeatmapGo
• Changed local memory of the DSP to FFT Hardware Accelerator (HWA) controlled by the Cortex R4F.Go
• Changed DSP to HWAGo
• Changed 1D FFT data is averaged then subtracted from each sample.to Once the active chirp time of the frame is complete, the interframe processing can begin, starting with static clutter removal. 1D FFT data is averaged across all chirps for a single Virtual Rx antenna. This average is then subtracted from each chirp from the Virtual Rx antenna.Go
• Added With N_s = Number of samples per chirp; N_c = Number of chirps; N_r = Number of recieve antennas; X_nr = Average sample for a single recieve antenna across all chirps; X_ncr = Single Chirp from a receive antennaGo
• Changed Single Chirp... to Samples of a Single Chirp...Go
• Changed Capon beam former to Capon beam formingGo
• Added The Capon BF algorithm is split into two components: 1) the Spatial Covariance Matrix and 2) Range-Azimuth Heatmap Generation. The final output is the Range-Azimuth heatmap with beamweights. This is passed to the CFAR algorithm.Go
• Added computationGo
• Deleted Let s(t) be the incoming waves after mixing to baseband. The sensor array signal to be processed is given by: X(t) = A(θ)s(t) + n(t). Where: A(θ) = (a(θ 1 ), …a(θ M )) is the steering matrix. a(θ) = (e j2πy 1 sin(θ) , …, e j2πy N sin(θ) ) is the steering vector. M is the number of angle bins. y n is the sensor position normalized by wavelength. The Capon BF approach is: θcapon = argminθ{trace(A(θ) × Rn-1 × A(θ)H }, where Rn is the spatial covariance matrix. Go
• Added naGo
• Added Then, the beamforming weights are calculated as Go
• Added with CFARGo
• Added image title CFAR and Elevation EstimationGo
• Added using the CFAR smallest of method.Go
• Added This output is combined with the point cloud produced during CFAR and Elevation Estimation, resulting in output for each point of:Go
• Added image title Doppler Estimation and Combination of ResultsGo
• Deleted which can be used by classification layers (currently not implemented in this example of a people counting application).Go
• Added range, azimuth, elevation, doppler, and SNR for each point in the point cloud. It outputs a list of tracked objects - each object has position, velocity, and acceleration in 3D Cartesian (X, Y, Z) space.Go
• Deleted performs target localization, and reports the results (a target list). therefore, the ourpur of the tracker is a set of trackable objects with certain properties (like position, velocity, physical dimensions, point density, and other features.Go
• Changed ...and memory consumption of the processing chain, up to and including the tracking. to ...consumption of the signal processing chain running on the DSP. Time remaining assmumes a 50 ms total frame time.Go
• Changed MIPS Use Summary table and added the following rows Range-Azimuth Heatmap Generation, 2 Pass CFAR, Elevation Estimation, Doppler Estimation, Total Processing Time, and Total TimeGo
• Changed Task Model to DPM ModelGo
• Deleted High-level processing is implemented with two tasks: higher priority mailbox task, and lower priority application task. When the system is configured, the mailbox task is pending on a semaphore, waiting for the frame ready message from DSP. When awakened, the mailbox task copies the relevant point cloud data from the shared memory into TCM, and posts the semaphore to an application task to run. It then creates the transport frame header, and initiates a DMA process for each part (TLV) of the frame. While DMA started sending data over UART, the mailbox task yields to the lower priority application task. When the DMA process completes, additional DMA can be scheduled (for example, TM2 and TM3). To achieve parallelism between the task processing and DMA, the transmit task sends the current (Nth) point cloud TLV with the previous (N-1)th target list and target index TLVs.Go
• Added within a DPUGo
• Changed The application task creates an algorithm instance...to The DPM initializes and configures the tracker DPU...Go
• Changed angle, Doppler... tto azimuth, elevation, Doppler, SNR...Go
• Added Each iteration of the tracker runs through the following steps: Go
• Changed The predict function estimates the...toIf a track exists, use a Kalman filter to predict the tracking...Go
• Changed The association function... to If a track exists, the association...Go
• Changed Points not assigned go through...to Any points not assigned to a track go through...Go
• Changed measurementto 3D Cartesian + DopplerGo
• Changed ...multiple tests to becometo a threshold for cumulative SNR and threshold for minimum number of points in the set to become a new track.Go
• Added list itemGo
• Added trackerEnabledGo
• Added 1Go
• Added -Go
• Added Enables or disables the tracker. 1 is enabled.Go
• Changed 250 to 1000Go
• Changed NA to 67Go
• Added * 10Go
• Changed NA to 105Go
• Added * 1000Go
• Added - dependent on chirp configuration Go
• Changed -1.5 to -6Go
• Changed -1.5 to 6Go
• Changed lower Entrance to Near EntrancesGo
• Changed 1 to 0.5Go
• Changed Entrance area lower boundary, in meters, set to ) if not defined to Wall/Entrance closest to EVM. The distance along the Y axis from the EVM where you want to start tracking.Go
• Changed upperEntrance to Far EntranceGo
• Changed 4.5 to 7Go
• Changed Entrance areas lower boundary, in meters, set to 100 if not defined.to Wall/Entrance far from the EVM. The distance along the Y Axis where you want to start/stop tracking. Go
• Added two rows to the table (Floor and Ceiling)Go
• Changed 100 to 0Go
• Added table row: SNR Obscured thresholdGo
• Changed 5 to 15Go
• Added If the target is in a static zone and is only associated with static points (0 Doppler), then the sleep2Free threshold is used. Go
• Added table row: sleep2freeThreGo
• Changed Volume to GainGo
• Changed 4 to 3Go
• Changed 3 to 1.5Go
• Changed 2 to 1.5Go
• Added table row: HeightLimitGo
• Changed 0 to 20Go
• Added The gating gain is a factor by which the gating volume can be increased to search for points to associate with the track. Go
• Changed range to lengthGo
• Added This lab can also run on the ODSISK or AOP module.Go
• Changed IWR6843 EVMto IWR6843ISK EVMGo
• Added This section will discuss the performance testing technique and results. During the development of the demo, TI developed a new technique to measure the accuracy of the tracker. As a result, the tracker accuracy values will be very different and much more detailed than previously published results for other TI People Counting demos. Go
• Deleted The people counting system was set up for five different test cases, to characterize and demonstrate the capabilities of the system.Go
• Deleted This test scenario shows the ability to change the detection and tracking range of the people counting system by modifying the chirp configuration used. A person is shown to be detected at 14 m, and the history of the person's movement is tracked as he moves through the scene.Go
• Added table column Number of PeopleGo
• Changed all 100's to NA in this tableGo
• Added From these results, we can see that the software performs well when the targets are moving. The static clutter removal algorithm discussed previously removes any noisy points generated from non-human objects that don't interest us. This leaves each person as a singular, isolated cluster of points. As we add more people to the scene, the tracker accuracy slowly decreases, due to complexities introduced in the point cloud when multiple people interact with eachother. The tracker performance is also poor when many people in the room are sitting, as they become static and are no longer detected by the point cloud. This can be rectified either through point cloud improvements, tracker improvements, or both. Otherwise, the people counting and tracking algorithm shows that the mmWave IWR6843 device is more than capable of tracking people in complex, indoor environments. Go

Changes from B Revision (November 2018) to C Revision

• Changed 57 GHz to 60 GHz in Section 0.1Go
• Changed 57 GHz to 60 GHz throughout Section 0.2.2.1Go
• Changed 7-GHz available bandwidth to 4-GHz available bandwidth in Section 0.2.2.1Go
• Changed 84 GHz to 64 GHz in Section 0.2.2.1Go
• Deleted NS - Number of samples per chirp;Go

Changes from A Revision (April 2018) to B Revision

• Changed all instances of IWR1642 to IWR6843Go
• Changed all instances of 77 GHz to 60 GHz in reference to the IWR6843Go
• Changed information in Go
• Added note regarding IWR1642 in 2T Mode.Go
• Changed information in Section 0.1Go
• Changed data in Table 1Go
• Changed information in Section 0.2.1.2.2Go
• Changed information in Section 0.2.2.1Go
• Changed data in Table 2Go

Changes from * Revision (March 2018) to A Revision