Building a self-balancing bot before next semester. Using MPU6050 for angle. Bench test, motor off: gyro drift accumulates fast, ~1 degree per 10 seconds.
Tried just gyro integration: drifts hard, falls within 30 sec.
Tried just accel: too noisy, balance impossible.
Combined with weighted average (0.98*gyro + 0.02*accel): better but still drifts slowly.
People recommend Kalman filter but math hurts brain. Complementary filter looks doable. Anyone implement complementary on AVR (Uno) with stable balance over 10+ minutes?
Side question: madami nakapagsabi na DMP (Digital Motion Processor) on MPU6050 already does this. Worth using vs writing my own filter?