In the calibration of robot tool coordinate system,the constraint equation was established by using the distance between any two points in the robot space to remain unchanged in the robot base coordinate system and the measuring instrument coordinate system.For solving such nonlinear problems of the equation,biased random key genetic algorithm (BRKGA) was introduced.In order to enhance the stability of algorithm and accelerate the convergence of algorithm,local search was added for improvement.The simulation results show that the solution accuracy of improved BRKGA is improved by 17% and 18% than the least square iteration and Newton method respectively.The results of physical experiments show that the solution accuracy of improved BRKGA is improved by 22% and 8% respectively.The solution accuracy variance of improved BRKGA is smaller than that of BRKGA,and the number of iterations is less than that of BRKGA while it converges,indicating that improved BRKGA has better stability and faster convergence.