edumip balance code filter coefficients

lusher00 · December 18, 2021, 1:05am

I was hoping the creator of rc_balance.c for the edumip robot could share the model used to determine these filter coefficients

github.com

beagleboard/librobotcontrol/blob/master/examples/src/rc_balance_defs.h

/**
 * rc_balance_defs.h
 *
 * Contains the settings for configuration of rc_balance.c
 */

#ifndef RC_BALANCE_CONFIG
#define RC_BALANCE_CONFIG

#define SAMPLE_RATE_HZ		100	// main filter and control loop speed
#define DT			0.01			// 1/sample_rate

// Structural properties of eduMiP
#define BOARD_MOUNT_ANGLE	0.49 // increase if mip tends to roll forward
#define GEARBOX			35.577
#define ENCODER_RES		60
#define WHEEL_RADIUS_M		0.034
#define TRACK_WIDTH_M		0.035
#define V_NOMINAL		7.4

This file has been truncated. show original

// inner loop controller 100hz
#define D1_ORDER 2
#define D1_NUM {-4.945, 8.862, -3.967}
#define D1_DEN { 1.000, -1.481, 0.4812}
#define D1_NUM_LEN 3
#define D1_DEN_LEN 3

// outer loop controller 100hz
#define D2_ORDER 2
#define D2_NUM {0.18856, -0.37209, 0.18354}
#define D2_DEN {1.00000, -1.86046, 0.86046}
#define D2_NUM_LEN 3
#define D2_DEN_LEN 3

I’ve made several balancing robots but I’ve never done it with this kind of filter and I would like to understand it a little better. How would I determine new numbers for something with a bit more mass?