Robotic avatar and telexistence systems have risen in prominence after the covid-19 pandemic, where current telecommunication methods are limited in terms of physical interaction abilities. Most existing systems focus on manual control of the remote robot, where the robot's arms and head movements follow the user's movements. Despite the effectiveness of such controls in conveying high levels of embodiment, such control methods jeopardize the efficiency of controls, especially for complex physical manipulation tasks, unclear environments, or unstable communication. Therefore, we propose an assistive-manipulation method to augment users' control of a telexistence robot during physical manipulation tasks. Machine Learning (ML) was used in the remote environment to localize target objects. This information is sent to the local environment where an inverse kinematic (IK) solution to hold the intended object is generated. The generated IK solution is fused with the one generated by the user's arm movements. The system enables generating various levels of IK fusion. However, an essential aspect of telexistence is to maintain high levels of embodiment and body ownership over the remote robot. Therefore, the evaluation in this paper focuses on investigating the effect of haptic feedback and the level of IK fusion on body ownership. The results indicate that haptic feedback induced a sense of assurance of task completion and enabling assistance from the system improved the user's sense of control over the robotic arm.