Usage

prepare

Request algorithm library and place the algorithm library to ${SDK}/vendor/amlogic/aml_csi/alg_lib/libaaisdk.so. And place related header files to following path: ${SDK}/vendor/amlogic/aml_csi/alg_lib/include/iva/iva_wisense_ppd.hpp ${SDK}/vendor/amlogic/aml_csi/alg_lib/include/iva/iva.hpp ${SDK}/vendor/amlogic/aml_csi/alg_lib/include/iva/iva_wisense_ppd.hpp

Custom led

${SDK}/vendor/amlogic/aml_csi/main.cpp adaptation wisense_led_init()& wisense_led_set() function.

Compile

Take ba401 as an example, add the following configuration to ${SDK}/buildroot/configs/amlogic/a4_ba401.config file:

BR2_PACKAGE_AML_CSI=y

run the following command to compile:

make aml-csi-rebuild && make

Config network

Start the system and then close system log(Optional)：

echo 1       4       1      7 > /proc/sys/kernel/printk

Connect to wifi：

wpa_cli -iwlan0 remove_network 0
wpa_cli -iwlan0 add_network 0
wpa_cli -iwlan0 set_network 0 ssid '"WiFi name"'
wpa_cli -iwlan0 set_network 0 key_mgmt WPA-PSK
wpa_cli -iwlan0 set_network 0 psk '"password"'
wpa_cli -iwlan0 set_network 0 pairwise CCMP TKIP
wpa_cli -iwlan0 set_network 0 group CCMP TKIP
wpa_cli -iwlan0 set_network 0 proto RSN
wpa_cli -iwlan0 enable_network 0
wpa_cli -iwlan0 status
wpa_cli -iwlan0 save
dhcpcd wlan0

Run aml_wisense demo

aml_wisense -l level

level represents the detection sensitivity level, and the specified parameters are 1, 2, 3, and 4. The larger of the number, the higher the detection sensitivity of the algorithm.

Development

Pipeline

         csi data              csi data              wi_out_ret

wifi driver ---------> aml_wisense --------> wisense alg ----------> result

parameter

aml_wisense will automatically obtain csi data and send it to the algorithm interface, and the algorithm will return the operation result(wi_out_ret).

wi_out_ret as follows:

• wi_out_ret.motion_status(int):

  • 1: There is motion that meets the level requirements.
  • 0: There is no motion that meets the level requirements.

• wi_out_ret.motion_class(int): The motion level detected by the algorithm, the larger of the value, the smaller the motion.

  • 1 or 2: Small motion, detect close range motion. A single person is within 3m range, and multiple people is within 5m range.
  • 3 or 4: Large motion, detect long distance motion or close range slight motion. Range up to 12m.
  • 0: There is no motion.

• wi_out_ret.motion_value(float): Motion complex value. Float value from 0 to 1, it reflects the complexity of the current motion environment. The larger of the value, the more complex of the current motion environment is.