Journal of Scientific Research and Reports

Agricultural robotic manipulators are increasingly required to execute accurate, rapid and low-damage motions within crop environments that are geometrically irregular, mechanically compliant and subject to uncertain sensing. The number and arrangement of degrees of freedom determine only part of their capability: the decisive engineering problem is how kinematics, rigid-body and contact dynamics, control, perception and task scheduling combine during real operations. This critical narrative review examines the dynamic modelling and control of multi-degree-of-freedom manipulators used principally for selective harvesting, with comparative attention to pruning, crop handling and related field interventions. Peer-reviewed engineering literature published from 2000 to 13 August 2026 was considered alongside foundational robotics research. Evidence was interpreted by distinguishing analytical modelling, controlled prototype testing and integrated field evaluation. The synthesis identifies three recurrent tensions. Additional joints may expand access and obstacle avoidance but increase calibration, dynamic coupling and control burden; aggressive motion profiles may shorten nominal execution time while increasing residual vibration, missed grasps or crop damage; and increasingly sophisticated planners and controllers may show advantages in simulation without corresponding evidence of superior whole-system productivity. Studies of hybrid manipulators, redundant fruit-picking arms, force-regulated grippers and coordinated multi-arm harvesters demonstrate feasible methods, yet few isolate the contribution of dynamics-aware control under comparable field conditions. Field evaluations further reveal marked effects of canopy structure, visibility, detachment technique and success-rate denominators. A task-centred design framework is advanced in which manipulator morphology, uncertainty estimation, torque- and contact-constrained trajectories, and crop-safe interaction are evaluated jointly. Priority research needs include validated plant–robot contact models, parameter identification under changing payloads, transparent energy and damage measurements, and multi-site comparisons against simpler control baselines. Greater model complexity should be justified by reproducible gains in useful, marketable agricultural output rather than movement accuracy alone.