The recent evolution of surgical robots has resolved a number of ergonomic issues associated with conventional minimally invasive surgery (MIS) in terms of aligned visiomotor axes, motion scaling and ergonomics. One of the latest advances is the introduction of human-robot cooperative control combining features such as active constraints, machine learning and automated movements. This paper aims to integrate these techniques into a framework which can be generalized to a wide range of surgical tasks. This paper proposes a system entitled Hubot; a Human-Robot collaborative framework which combines the strengths of the surgeon, the advantages of robotics and learning from demonstration into a single system. Hubot was successfully implemented on a Raven II surgical robot and a user study was conducted to evaluate its performance. Both a training and a simulated clinical case were investigated and showed promising results in comparison to fully manual task execution, including reduced completion time, fewer movements for the operator and improved efficiency.