本文是 Nordic nRF54LM20-DK 物联网开发套件系列教程的第 6 篇内容。
本文承接第 5 篇手势识别演示教程的内容继续讲解。本文将演示如何编译固件，以实现从加速度传感器与陀螺仪传感器采集数据。该功能一般用于采集数据，用以训练全新的模型。若要在 minicom 串口终端中采集数据，需要修改下述 prj.conf 配置文件，在文件中添加如下配置参数：
CONFIG_DATA_COLLECTION_MODE=y
CONFIG_BLE_MODE_NONE=y
该配置文件路径如下：
edge-ai/applications/gesture_recognition/configuration/nrf54lm20dk_nrf54lm20b_cpuapp/prj.conf
随后按照第 5 篇教程的步骤，编译并烧录程序。若操作全部正常完成，在独立的 minicom 串口终端中，将会输出如下日志信息：
*** Booting MCUboot v2.3.0-dev-3dfaa012cf34 ***
*** Using nRF Connect SDK v3.3.0-preview2-ede152ec210b ***
*** Using Zephyr OS v4.3.99-4b6df5ff11b1 ***
I: Starting bootloader
I: Primary image: magic=unset, swap_type=0x1, copy_done=0x3, image_ok=0x3
I: Secondary image: magic=unset, swap_type=0x1, copy_done=0x3, image_ok=0x3
I: Boot source: none
I: Image index: 0, Swap type: none
I: Bootloader chainload address offset: 0x10000
I: Image version: v0.0.0
I: Jumping to the first image slot
*** Booting MCUboot v2.3.0-dev-3dfaa012cf34 ***
*** Using nRF Connect SDK v3.3.0-preview2-ede152ec210b ***
*** Using Zephyr OS v4.3.99-4b6df5ff11b1 ***
I: Starting bootloader
I: Primary image: magic=unset, swap_type=0x1, copy_done=0x3, image_ok=0x3
I: Secondary image: magic=unset, swap_type=0x1, copy_done=0x3, image_ok=0x3
I: Boot source: none
I: Image index: 0, Swap type: none
I: Bootloader chainload address offset: 0x10000
I: Image version: v0.0.0
I: Jumping to the first image slot
*** Booting nRF Connect SDK v3.3.0-preview2-ede152ec210b ***
*** Using Zephyr OS v4.3.99-4b6df5ff11b1 ***
[00:00:00.048,787] <inf> main: nRF Edge AI Gestures Recognition Demo:
[00:00:00.048,798] <inf> main: nRF Edge AI Runtime Version: 2.2.0
[00:00:00.048,816] <inf> main: nRF Edge AI Lab Solution id: 36038
0,0,0,0,0,0
0,0,0,0,0,0
0,0,0,0,0,0
5904,-351,-7637,0,0,0
5953,-332,-7722,39,42,-17
5948,-326,-7720,35,33,-17
5912,-313,-7714,29,29,-11
5912,-314,-7703,15,23,-7
5935,-302,-7705,2,13,-4
5937,-314,-7709,-10,4,-1
5934,-335,-7708,-23,-4,2
5921,-365,-7705,-30,-15,6
5937,-384,-7699,-35,-21,11
5953,-408,-7695,-33,-29,13
5950,-426,-7702,-29,-37,13
5955,-430,-7696,-22,-42,20
5978,-444,-7697,-14,-44,29
5990,-429,-7698,-5,-44,30
6004,-408,-7692,2,-44,27
6015,-409,-7689,9,-39,23
6029,-387,-7695,14,-31,20
6033,-368,-7688,14,-22,13
6029,-350,-7676,13,-14,7
6029,-347,-7698,12,-5,4
6020,-343,-7692,6,6,4
6035,-348,-7694,1,17,1
6034,-347,-7690,-2,24,-3
6009,-355,-7689,-7,33,-11
6001,-373,-7698,-7,39,-20
5983,-386,-7716,-6,44,-28
5967,-408,-7697,-2,44,-35
5961,-396,-7691,2,42,-30
5916,-339,-7680,4,37,-13
5984,-286,-7771,-9,20,29
5903,-330,-7692,-35,39,37
5983,-337,-7664,-24,5,-10
5925,-360,-7721,-24,21,-31
5930,-402,-7674,-30,2,-16
5938,-411,-7744,-30,2,8
5936,-432,-7677,-28,-15,16
5941,-421,-7719,-12,-22,10
5950,-429,-7703,-3,-31,13
etc
etc
.....
以上数据为三轴加速度、三轴陀螺仪惯性测量单元（IMU）的原始采集数据，采样频率为 100Hz（忽略前 4 组采样数据）。
<acc_x>,<acc_y>,<acc_z>,<gyro_x>,<gyro_y>,<gyro_z>
传感器采用DigiKey在售的Bosch BMI270 惯性测量单元 Sparkfun 评估板。
上述 Zephyr 实时操作系统的配置参数，会启用send_imu_data(input_data_i16)函数，将传感器数据如上所示输出至串口。同时该模式下，低功耗蓝牙（BLE）功能会被关闭。
敬请关注本系列后续文章。本文讲解了如何借助 Axon NPU 驱动，直接在这款创新 Nordic 物联网平台上，为手势演示案例采集模型训练数据。这款斩获 2026 嵌入式世界展会奖项的 Nordic nRF54LM20-DK 物联网开发套件，现已在DigiKey上架开售。