Abstract: To address the challenge of simultaneously maintaining operator intent, ensuring safe obstacle avoidance, and balancing multiple motion modes in remote control of amphibious unmanned platforms in complex environments, a multimodal human-machine collaborative path planning and trajectory tracking method was proposed. Based on ground-based tracked and airborne ducted flight dynamics models, motion primitives were used to structurally express operator input and unified optimization and tracking of the reference path was achieved within a Model Predictive Contour Control (MPCC) framework. The method was validated using the Gazebo–PX4 co-simulation platform in a complex environment containing narrow passages and windows. Results show that the platform achieved a minimum safe clearance of 0.32 m in ground mode and stably passed through a 0.8 m × 0.8 m narrow window with a minimum safe clearance of 0.24 m in flight mode. Furthermore, both motion modes achieved short task completion times and smooth, controllable speed output, validating the effectiveness of the proposed method in complex and constrained environments.