Robotic teleoperation systems enable humans to control robots remotely. Recent advancements in VR have transformed teleoperation into immersive, intuitive platforms that improve human-machine synergy. However, existing VR-based teleoperation systems face challenges such as under-informed shared control, limited tactile feedback, and computational inefficiencies, which hinder their effectiveness in complex, cluttered scenarios. This paper introduces a novel VR-based teleoperation system incorporating haptic feedback and adaptive collision avoidance. The system tracks human hand trajectories via an VR-based handheld device and provides tactile feedback from a robotic gripper. Through the proposed tele-MPPI method, the system anticipates robot motions, adaptively adjusts trajectories to avoid obstacles, and maintains computational efficiency, enabling real-time operation at 20 Hz. Simulations and experiments demonstrate the system’s ability to grasp delicate objects without damage and to navigate cluttered environments by mitigating collision risks.