Quadruped robot- "dog"

Hello everyone, last two months I am making one robot with four legs. And until now I can sit, stand up etc, everything with all four legs on the ground- it is normal, but when it walks it is very unstable. I thinking about how to make the walking algorithm and stability control, but have no idea how. I would be very happy to hear something useful from you.

My project is with Arduino Mega, PCA9685, 12- MG995R and MPU6050. I can read the date from gyro but cant find a way to use it.( only when it on four legs and the ground change angle ) I have very basic inverse kinematics algorithm to calculate the legs separatly, so I give (x,y,z) to the legs, but not all together and with the torso. I have no simulation program and I dont know how to use some from the popular. I am new in the robotics as all and it is my first project, it is difficult, I know, but would try.

Thank you very much!

You are trying to go from simple movements with multiple feet on the ground, to balancing on 2 legs? I am pretty sure that is a monstrously huge leap in complexity. And this is part of the reason I played with hexapods. It was easy to make a walking gait that always had 3 feet on the ground.

Hello, thank you that you asking! I tried "2 steps walking"( where it had to balance on 2 legs, but it so, as you said), also "4 steps walking" but it is very slowly and isnt so stable too) Imagine that I put camera on top, it wont be nice video. And yes, I make simple movements- Leg 1=>position 1, torso=> position 2( I made one new update to my code and now I can move the torso in different directions with all legs together) , Leg 2=> position 3 etc..

If you use your favorite search engine together with the phrase "walking algorithms", many interesting discussions will turn up! Walking is very complex behavior.

Thank you! I searched , but didnt find enough info for me to start updates.

More specifically, "quadruped walking gait" (a search suggestion that comes up when you enter "walking algorithms") turns up 336,000 hits.

Here is the first page of the result for images. It is a heavily studied and extremely well documented topic.

|500x401

Capture.PNG|1118x898

Thank you very much! I will see.

Thank you again! I found a bit more information after your advice, but still lots to discover.