Underwater manipulation with free-floating autonomous underwater vehicles (AUVs) is an under-explored research area that this paper addresses. The open-source mechanical, electrical, and software designs of an AUV and continuum manipulator system are provided as a platform for performing this research. The underwater robot system has high degrees of freedom including the vehicle body motion and the manipulator joints. Therefore, when performing a manipulation task, the robot has many different potential trajectories which satisfy the task constraints, and this kinematic redundancy needs to be resolved. This paper provides a method for solving the redundancy problem. The relevant kinematic models are derived in order to build an algorithm to calculate desired joint velocities in real time. Different methods to optimize the algorithm for specific tasks are proposed, including a basic weighting method and a gradient projection method to optimize a user-defined objective function. Both simulation and experimental results are analyzed to assess the performance of this algorithm.

翻译：针对自由漂浮自主水下航行器（AUVs）的水下操作这一尚未充分探索的研究领域，本文提出了一种解决方案。我们提供了开源AUV与连续体机械臂系统的机械、电气及软件设计，作为开展该研究的平台。该水下机器人系统具有高自由度，包括载体运动与机械臂关节动作。因此，在执行操作任务时，机器人存在多条满足任务约束的潜在轨迹，需解决此类运动学冗余问题。本文提出一种冗余问题的解析方法：通过推导相关运动学模型构建实时计算期望关节速度的算法，并针对特定任务提出多种优化策略，包括基础加权法及梯度投影法用于优化用户自定义目标函数。通过仿真与实验数据分析评估了该算法的性能。