Introduction Reducing treatment duration in proton therapy is one of the key factor to increase the number of patient treated per day. To overcome this issue, we have developed novels haptic manual guidance control implemented on the patient positioning robot (Orion PT) to be used during the pre-positioning phase by the paramedical assistant. These manual control modes allow the paramedical assistant to be close to the patient while reducing the time for the pre-positioning setup. The four manual control modes of the robot are compared with a generic reach-and-point trajectory. Methods The main objective of this article is to compare intuitiveness of the four following manual guidance modes on the proton therapy robot already presented in [1] : - TeachPendant remote sensor with one button per Cartesian axis (with a total of 6). This is the reference control mode used in actual radio/proton therapy treatment room. - U bar haptic device located under the couch top and mounted on Force/Torque sensor to follow paramedical assistant intention in measuring applied force. This control mode was previously presented in [2] . - Couch top comanipulation mode. Paramedical assistant is able to move the couch top directly in applying force upon it. A F/T sensor located between couch top and robot end-effector measure operator intentions and move the couch top accordingly. The benefits of the haptic guidance compared to the previous one is the ability for the paramedical assistant to manipulate the robot directly from the couch top from any location around it, increasing intuitiveness. - 3D haptic joystick with translations as well as rotations placed on the couch top. The robot move as the same way as the 3D joystick does. A comparison between those haptic control modes will be done using a reference trajectory and the following criterion will be used to assess the intuitiveness of the proposed methods: - Movement duration to reach target. - Mean error between real and reference trajectory. - Learning curve. Results By now, the fourth haptic devices can be used with the robot and have been successfully tested. Several quantitative tests show intuitiveness improvement with U-bar haptic device compared to traditional control with teachPendant. Some qualitative tests will be carry out to measure the benefit of the three manual haptic control of the robot previously described. Conclusions This paper shows three novels haptic device to manually control a proton therapy robot during pre-positioning phase performed by paramedical assistant. Qualitative tests shows benefit of one of the proposed methods and quantitative tests will be executed soon.