Robot Obstacle Avoidance

Hi guys,
I’m working in a project on robotics, and i want to know if there is a possibility to detect obstacle and avoid them with the OpenMV cam M7 ?

Thaaanks,

Yes, what computer vision method do you plan to use. I.e. how do you envision seeing the object? Typically there needs to be a feature you are looking for.

Thanks a lot mr kwagyeman for you reply,
I want to make an intelligent transport system and for that i need that my robot can detect obstacles or an other robot to avoid them .
for that i want to use Zumo robot and the openmv cam m7 a servo motor.

the problem here is firstly i don’t know how to link the camera to the robot…the shematic wrinig diagram

It’s something like the project GitHub - mjs513/TeensyOpenMV but i don’t know when and where to start.

can you help please sir.

Thanks,

Ah, yeah, you can actually do everything from the OpenMV Cam.

How are the motors driven? Servos? Or are they two wires? If two wires then please buy our motor shield. If they are servo motors please buy our servo shield. The Pan and Tilt shield will allow control of 3 motors and it has a regulator to take battery power as an input.

Thank you sir,

okey i will buy them after coronavirus, but if there is another techniques or something to use to do this project just with openmv cam and servo motors of openmv cam i will be really to do the project this days i will be thankfull.

Okay, if you have servo motors then you can use pins 7, 8, 9 on the OpenMV Cam H7 to control 3 servos. Please see the board control example scripts.

Once you have motor control working you can do other stuff.

Okay thaaaaaaaaanks a loot :wink:

On the same theme. I have build a fully autonomous outdoor robot (differential drive) and I use lidar and sonar for avoidance. I will buy and openMV H7 plus and use it to build a tree and wooden post database for avoidance. I am new to this area so would appreciate any tips. Regarding training I want to be able to trigger the avoidance about 3m from the tree or post. The robot moves at 0.5m/s so we have plenty of time to avoid the tree or post. I know I have to take many pictures of tree trunks to make the learning useful. In you video you took 100 smiley pics and 100 frowns. For a tree/post project what do you suggest regarding volume and type of pictures for this project. Would it be better to do this with thermal? Strong daylight vs early morning
Also any technical issues that you can see
Do you know of an already existing project
Sorry to ask so many questions
Max

The OpenMV Cam is not a distance sensor. You should really use a distance sensor for object avoidance.

Anyway, just take in general a very large number of pixels of the trunk and not the truck. Aim for like 1K pics of each with different lighting conditions.

You need to keep adding data to your model until your training results show it segmenting the class well.