Jetbot
Jetbot is an official robotics platform supported by nVidia for their Jetson Nano series of embedded systems.
Waveshare Jetbot
The
Waveshare JetBot was the most cost effective version of a Jetbot kit to purchase in Australia.
Waveshare Wiki is out of date
As at 28 December 2020, the Waveshare Wiki contains errors.
The official Waveshare Wiki for Jetbot is found at:
https://www.waveshare.com/wiki/JetBot_AI_Kit
Error #1
The link to the Jetpack SD Card image is out of date. It points to version 4.3 of Jetpack. Version 4.4.1 is the official current version of Jetpack and is compatible with the Waveshare kit (see Error #2 below).
Error #2
There is no need to undertake step 5 and clone Waveshare's github, the official NVidia Jetbot software supports the Waveshare board "out-of-the-box".
which supports the Waveshare board.
The Waveshare OLED is also supported by the official Jetbot software.
Using Official Jetpack 4.4.1 images
The official Jetpack image for Jetbot is version 4.4.1 and is 8.8Gb in size. nVidia host it on Google Drive and I have found Chrome is the best browser for actually downloading large files from Google Drive.
Powering the Board
Waveshare have not documented the fact that even with fully charged 18650s installed, you will need to use the 12.6V charger that came with the kit before the Waveshare Jetbot board will turn on.
It only takes a small amount of time with the 12.6V charger connected.
Similarities with the Original Jetbot
The original Jetbot hardware leverages the Adafruit Raspberry Pi MotorHat board and the PiOLED.
The Waveshare board has some similar and some different hardware components
Original
Motor Driver: TB612
PWM signal generator: PCA9685 (I2C)
OLED Controller: I2C
Waveshare
Motor Driver: DVR8870 (earlier revision of the current board)
PWM signal generator: PCA9685 (I2C)
OLED Controller: I2C
Battery Protection: S-8254AA
DC Buck Convertor: APW7313
Power Supply
It is worth noting that the Waveshare board can only supply 3A and thus you must put your Jetson Nano in 5W mode when running, powered by the Waveshare board.
Initial thoughts
It is an excellent kit with some good documentation but due to not allowing public editing of the Waveshare wiki, the included documentation is out of date and a little lacking.
The choice of a 3A power supply is disappointing as this requires you to turn off 2 of the 4 cores in software when running. It would have been nice to include a 4A solution and allow all 4 cores to run.
[Update 30 Dec 2020: In testing I have not seen power consumption with MAXN enabled that exceeds 10Ws, so all 4 cores can be enabled when running from battery]
Update 29 Dec 2020 - Waveshare Jetbot runs too fast...and the fix
So it would appear that when using the currnt nVidia Jetbot code, the collision_avoidance demo runs the motors extremely quickly.
Seven days ago, Waveshare push an update to their GitHub that appears to correct this, the relevant commit is:
https://github.com/waveshare/jetbot/commit/8dc6cf65899fb0e7691f8d0ec38daaee63970807
Note: This has been integrated with the official nVidia Jetson Nano
Update 30 Dec 2020
A test of the latest Github code for collision avoidance:
Update 16 Jan 2021
The usable speed values for the Jetbot code with the Waveshare kit can be found at the Jetbot Github here.
Click to see these values in action
Update 15 July 2021
The speed values above are a hack. The real problem is that the motors are receiving 12 volts when they should receive 5 or 6 volts.
User Chitoku on GitHub corrected the hardware and detailed his apparent results in this post:
https://github.com/NVIDIA-AI-IOT/jetbot/discussions/377
The h-bridge motor driver used by WaveShare is this one:
https://pdf1.alldatasheet.com/datasheet-pdf/view/807693/TOSHIBA/TB6612FNG.html
Jetbot
Comments