The objective of this thesis was to examine the interaction between user safety and cognitive-motor performance during reaching movements executed with a robotic arm through a human body machine interface (HBMI). Specifically, the effects of a safety controller on user cognitive workload and kinematics were assessed during learning the control of a simulated prosthetic arm through limited head movements. The results revealed that, compared to the group performing without the safety controller, the users assisted with the safety controller exhibited: i) a lower rate of information transfer, ii) a higher cognitive workload and iii) a reduced number of times the user brought the robotic arm close to the workspace boundaries when performing the adaptive reaching task. These results suggest that the autonomous safety controller increased user cognitive workload and reduced information transfer but provided a safer environment. This work contributes to the development of assistive technology such as HBMI and neuroprosthetics.