Robot Simulation (Kinematics)

6DOF robot kinematics in JavaScript.

Install

npm install robot-simulation --save

Use

import RobotSimulation from 'robot-simulation'

const geometry = [
      [1,  1,  0], // V0: 1x 1y
      [0, 10,  0], // V1: 10y
      [5,  0,  0], // V2: 5x
      [3,  0,  0], // V3: 3x
      [0, -3,  0], // V4: -3y
    ]

const RobotKin = new RobotSimulation(geometry)

let angles = [1.57, 1.2, 0, 0.3, 2.2, 1.1]

const pose = RobotKin.forward(...angles)[5]

angles = RobotKin.inverse(...pose)

Geometry

The geometry array consists of 5 entries describing the links V0-V5. Each Vn is a tuple of 3 coordinates from Jn to Jn+1. One constraint: The y,z of V3 and x,z of V4 must be 0 for the kinematics to work.

API

forward

RobotKin.forward(R0, R1, R2, R3, R4, R5)

returns

[ 
  [  0,      0,     0 ], //J0
  [  0.5,    1,  -0.8 ], //J1
  [ -0.2, -8.8,   0.3 ], //J2
  [  1.8, -5.6,  -2.8 ], //J3
  [  3.0, -3.6,  -4.7 ], //J4
  [  4.7, -1.3,  -5.5,  1,  6,  -2.8 ] //J5 + TCP Euler angles
]

inverse

X,Y,Z coordinates, A,B,C Euler angles in order 'abc'.

RobotKin.inverse(X, Y, Z, A, B, C)

returns

[  2, 1.6,  2.1, -3.5,  1, -1.5 ] //array of angles
[  1, 2.3,  3.1,  NaN, NaN, NaN ] //NaN for out of reach angles

kinematic coupling

This library assumes a robot with a series of joints. Some robots may have different kinematics. The depicted robot has a hinge at J1 and R1/R2 are at the same kinematic position. Therefore moving R1 also changes the angle at J2. To account for that, R2 has to move the same amount.

Using that information, you can use the solver to calculate the initial angles and correct them according to your kinematics.

let angles = RobotKin.inverse(...pose)
angles[2] += angles[1]
//set angles, do stuff 🤖

TODO

• robot configuration
• comply with DH for TCP orientation?
• more kinematic chains

Demo

See the kinematics in action: https://ulikgames.github.io/robot-control-panel/ The gui is available in this repo: https://github.com/UlikGames/robot-control-panel