your website name here

Research Activities

Intrinsically stable MPC for humanoid gait generation

The generation of Center of Mass (CoM) trajectories for humanoid walking requires solving unstable dynamics while satisfying stringent constraints in order to avoid falling. Model Predictive Control (MPC) has proven to be a very power tool for achieving this, but lacked formal stability. In this paper we introduce a way of introducing a stability constraint in a Quadratic Programming (QP) formulation. We also formulate a gait generation algorithm by including automatic footstep placement, so to provide a system able to track a reference velocity without need of specifying the position of the footsteps.
More information about this topic can be found here.
Real-time planning and execution of evasive motions for a humanoid Robot

A humanoid is standing in a workspace when a moving obstacle (such as a human) enters its safety area and heads towards it. The robot must plan and execute a fast evasive motion to prevent a collision. On the one hand, the possibility of taking steps makes this kind of obstacle avoidance possible; on the other hand, a suitable motion must be generated in real-time.

This behavior is the basic safety layer that a humanoid must have, in order to be able to share the same environment with humans. This work is inspired by the ongoing research project COMANOID which targets the deployment of humanoid robots in aeronautic assembly operations.
More information about this topic can be found here.
Task-oriented whole-body planning for humanoids

Consider a humanoid robot whose goal is to accomplish a given task (e.g. opening a door), possibly requiring stepping, in an environment cluttered by obstacles. This problem is interesting for many reasons: first, a humanoid robot has an high number of degrees of freedom; second, it is not a free-flying system, then its motions have to be generated such that the robot maintains the equilibrium.
We propose a probabilistic planner that is able to accomplish the given task while maintaining the robot equilibrium. In addition, our planner generates simoultaneously the footsteps and the whole-body motions to fulfil the above-mentioned goals. More information about this topic can be found here.

The planner is then extended by replacing the foot displacements with the movements of the center of mass. This makes the planner more general and enables to handle a wider variety of scenarios. More information about this topic can be found here.
Localization of an heterogeneous robot team

Consider a team composed by an aerial vehicle (UAV, equipped with sonar, camera and IMU) and several ground robots (UGVs, equipped with odometers) whose aim is to accomplish several tasks. Here, we address also the problem of anonymous measurements, i.e. the association between the UGV identities and camera measurements. A filter called ID-PHD reconstructs the UGV identities and provides estimates of the relative poses. A visual task is then considered in order to keep the UGV inside the camera field of view. Finally, formation control, navigation and obstacle-avoidance tasks are considered in order to safety navigate in an uknown environment.
More information about this topic can be found here.