Multi-objective trajectory planning for segment assembly robots using a B-spline interpolation- and infeasible-updating non-dominated sorting-based method

Abstract

The rapid and smooth functioning of segment assembly robots, which is always conflicting, is critical to improving efficiency and ensuring safety during tunneling construction, particularly for the series-actuated robots employed in non-circular shield machines. However, the trade-off between the aforementioned goals has not been explored for trajectory planning in joint space. Less is known about how to acquire superior trade-off Pareto solutions for this constrained multi-objective optimization problem. To fill this gap, this paper proposes a B-spline interpolation- and infeasible-updating non-dominated sorting-based method for multi-objective trajectory planning of shield machine segment assembly robots. In particular, a multi-objective optimization model is detailed in terms of time, acceleration, and jerk of joint trajectories while accounting for operational efficiency and motion smoothness. The given objective functions can be determined using B-spline interpolation and time information based on the known via-points of hydraulic joints. Meanwhile, the constraints are transformed into a limited number of control point-derived forms. Following that, the infeasible-updating non-dominated sorting-based evolutionary algorithm (INSEA) is introduced to solve this problem and find Pareto-optimal solutions. The main improvement is that the multi-objective function information of infeasible solutions is utilized so as to update the non-dominated sorting process, which adaptively applies population division and individual replacement. The findings indicate that the proposed method is capable of executing multi-objective trajectory planning during different stages of the assembly procedure, and the computed metrics are all greater than serval advanced algorithms. Furthermore, on the basis of the proposed INSEA, multi-degree B-spline interpolation provides lower acceleration peaks and smoother global trajectories than common cubic spline curves throughout the process. Therefore, this method can provide researchers with a wide range of alternatives to achieve the optimum trade-off for multi-objective trajectory planning.