Lorenzo Amatucci
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Lorenzo Amatucci has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Claudio Semini
Istituto Italiano di Tecnologia
5 shared publications - Victor Barasuol
Istituto Italiano di Tecnologia
4 shared publications - Giulio Turrisi
Dynamic Legged Systems Laboratory, Istituto Italiano di Tecnologia (IIT), Genova, Italy
3 shared publications - Angela Dai
Technical University of Munich
1 shared publication - Angelo Bratta
Dynamic Legged Systems Lab Istituto Italiano di Tecnologia (IIT) Genova Italy
1 shared publication - Dominique Orban
GERAD and Department of Mathematics and Industrial Engineering, Polytechnique Montréal, QC, Canada
1 shared publication
Latest Publications
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (12) , 14812-14819
ZiMPedance: Impedance-Aware ZMP Modeling and Control for Payload Carrying With Quadruped Robots
- 1 December 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (12) , 14812-14819
ZiMPedance: Impedance-Aware ZMP Modeling and Control for Payload Carrying With Quadruped Robots
- 14 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (1) , 1010-1017
Primal-Dual iLQR for GPU-Accelerated Learning and Control in Legged Robots
This letter introduces a novel Model Predictive Control (MPC) implementation for legged robot locomotion that leverages GPU parallelization. Our approach enables both temporal and state-space parallelization by incorporating a parallel associative scan to solve the primal-dual Karush–Kuhn–Tucker (KKT) system. In this way, the optimal control problem is solved in $\mathcal {O}(\log ^{2}(n)\log {N} + \log ^{2}(m))$ complexity, instead of $\mathcal {O}(N(n + m)^{3})$, where $n$, $m$, and $N$ are the dimension of the system state, control vector, and the length of the prediction horizon. We demonstrate the advantages of this implementation over two state-of-the-art solvers (acados and crocoddyl), achieving up to a 60% improvement in runtime for Whole Body Dynamics (WB)-MPC and a 700% improvement for Single Rigid Body Dynamics (SRBD)-MPC when varying the prediction horizon length. The presented formulation scales efficiently with the problem state dimensions as well, enabling the definition of a centralized controller for up to 16 legged robots that can be computed in less than 25 ms. Furthermore, thanks to the JAX implementation, the solver supports large-scale parallelization across multiple environments, allowing the possibility of performing learning with the MPC in the loop directly in GPU.
- 14 November 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 11 (1) , 1010-1017
Primal-Dual iLQR for GPU-Accelerated Learning and Control in Legged Robots
This letter introduces a novel Model Predictive Control (MPC) implementation for legged robot locomotion that leverages GPU parallelization. Our approach enables both temporal and state-space parallelization by incorporating a parallel associative scan to solve the primal-dual Karush–Kuhn–Tucker (KKT) system. In this way, the optimal control problem is solved in $\mathcal {O}(\log ^{2}(n)\log {N} + \log ^{2}(m))$ complexity, instead of $\mathcal {O}(N(n + m)^{3})$, where $n$, $m$, and $N$ are the dimension of the system state, control vector, and the length of the prediction horizon. We demonstrate the advantages of this implementation over two state-of-the-art solvers (acados and crocoddyl), achieving up to a 60% improvement in runtime for Whole Body Dynamics (WB)-MPC and a 700% improvement for Single Rigid Body Dynamics (SRBD)-MPC when varying the prediction horizon length. The presented formulation scales efficiently with the problem state dimensions as well, enabling the definition of a centralized controller for up to 16 legged robots that can be computed in less than 25 ms. Furthermore, thanks to the JAX implementation, the solver supports large-scale parallelization across multiple environments, allowing the possibility of performing learning with the MPC in the loop directly in GPU.
- 19 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (5) , 4620-4627
MUSE: A Real-Time Multi-Sensor State Estimator for Quadruped Robots
This paper introduces an innovative state estimator, MUSE (MUlti-sensor State Estimator), designed to enhance state estimation's accuracy and real-time performance in quadruped robot navigation. The proposed state estimator builds upon our previous work presented in [1]. It integrates data from a range of onboard sensors, including IMUs, encoders, cameras, and LiDARs, to deliver a comprehensive and reliable estimation of the robot's pose and motion, even in slippery scenarios. We tested MUSE on a Unitree Aliengo robot, successfully closing the locomotion control loop in difficult scenarios, including slippery and uneven terrain. Benchmarking against Pronto [2] and VILENS [3] showed 67.6% and 26.7% reductions in translational errors, respectively. Additionally, MUSE outperformed DLIO [4], a LiDAR-inertial odometry system in rotational errors and frequency, while the proprioceptive version of MUSE (P-MUSE) outperformed TSIF [5], with a 45.9% reduction in absolute trajectory error (ATE).
- 19 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (5) , 4620-4627
MUSE: A Real-Time Multi-Sensor State Estimator for Quadruped Robots
This paper introduces an innovative state estimator, MUSE (MUlti-sensor State Estimator), designed to enhance state estimation's accuracy and real-time performance in quadruped robot navigation. The proposed state estimator builds upon our previous work presented in [1]. It integrates data from a range of onboard sensors, including IMUs, encoders, cameras, and LiDARs, to deliver a comprehensive and reliable estimation of the robot's pose and motion, even in slippery scenarios. We tested MUSE on a Unitree Aliengo robot, successfully closing the locomotion control loop in difficult scenarios, including slippery and uneven terrain. Benchmarking against Pronto [2] and VILENS [3] showed 67.6% and 26.7% reductions in translational errors, respectively. Additionally, MUSE outperformed DLIO [4], a LiDAR-inertial odometry system in rotational errors and frequency, while the proprioceptive version of MUSE (P-MUSE) outperformed TSIF [5], with a 45.9% reduction in absolute trajectory error (ATE).
- 18 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (2) , 1265-1272
Non-Gaited Legged Locomotion With Monte-Carlo Tree Search and Supervised Learning
Legged robots are able to navigate complex terrains by continuously interacting with the environment through careful selection of contact sequences and timings. However, the combinatorial nature behind contact planning hinders the applicability of such optimization problems on hardware. In this work, we present a novel approach that optimizes gait sequences and respective timings for legged robots in the context of optimization-based controllers through the use of sampling-based methods and supervised learning techniques. We propose to bootstrap the search by learning an optimal value function in order to speed-up the gait planning procedure making it applicable in real-time. To validate our proposed method, we showcase its performance both in simulation and on hardware using a 22 kg electric quadruped robot. The method is assessed on different terrains, under external perturbations, and in comparison to a standard control approach where the gait sequence is fixed a priori.
- 18 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Robotics and Automation Letters
- Vol. 10 (2) , 1265-1272
Non-Gaited Legged Locomotion With Monte-Carlo Tree Search and Supervised Learning
Legged robots are able to navigate complex terrains by continuously interacting with the environment through careful selection of contact sequences and timings. However, the combinatorial nature behind contact planning hinders the applicability of such optimization problems on hardware. In this work, we present a novel approach that optimizes gait sequences and respective timings for legged robots in the context of optimization-based controllers through the use of sampling-based methods and supervised learning techniques. We propose to bootstrap the search by learning an optimal value function in order to speed-up the gait planning procedure making it applicable in real-time. To validate our proposed method, we showcase its performance both in simulation and on hardware using a 22 kg electric quadruped robot. The method is assessed on different terrains, under external perturbations, and in comparison to a standard control approach where the gait sequence is fixed a priori.
- Open Access
- 29 April 2024
- Published by Wiley in Journal of Field Robotics
- Vol. 41 (6) , 1829-1842
VERO: A vacuum‐cleaner‐equipped quadruped robot for efficient litter removal
Litter nowadays presents a significant threat to the equilibrium of many ecosystems. An example is the sea, where litter coming from coasts and cities via gutters, streets, and waterways, releases toxic chemicals and microplastics during its decomposition. Litter removal is often carried out manually by humans, which inherently lowers the amount of waste that can be effectively collected from the environment. In this paper, we present a novel quadruped robot prototype that, thanks to its natural mobility, is able to collect cigarette butts (CBs) autonomously, the second most common undisposed waste worldwide, in terrains that are hard to reach for wheeled and tracked robots. The core of our approach is a convolutional neural network for litter detection, followed by a time‐optimal planner for reducing the time needed to collect all the target objects. Precise litter removal is then performed by a visual‐servoing procedure which drives the nozzle of a vacuum cleaner that is attached to one of the robot legs on top of the detected CB. As a result of this particular position of the nozzle, we are able to perform the collection task without even stopping the robot's motion, thus greatly increasing the time‐efficiency of the entire procedure. Extensive tests were conducted in six different outdoor scenarios to show the performance of our prototype and method. To the best knowledge of the authors, this is the first time that such a design and method was presented and successfully tested on a legged robot.
- 29 April 2024
- Published by Wiley in Journal of Field Robotics
- Vol. 41 (6) , 1829-1842
Open AccessVERO: A vacuum‐cleaner‐equipped quadruped robot for efficient litter removal
Litter nowadays presents a significant threat to the equilibrium of many ecosystems. An example is the sea, where litter coming from coasts and cities via gutters, streets, and waterways, releases toxic chemicals and microplastics during its decomposition. Litter removal is often carried out manually by humans, which inherently lowers the amount of waste that can be effectively collected from the environment. In this paper, we present a novel quadruped robot prototype that, thanks to its natural mobility, is able to collect cigarette butts (CBs) autonomously, the second most common undisposed waste worldwide, in terrains that are hard to reach for wheeled and tracked robots. The core of our approach is a convolutional neural network for litter detection, followed by a time‐optimal planner for reducing the time needed to collect all the target objects. Precise litter removal is then performed by a visual‐servoing procedure which drives the nozzle of a vacuum cleaner that is attached to one of the robot legs on top of the detected CB. As a result of this particular position of the nozzle, we are able to perform the collection task without even stopping the robot's motion, thus greatly increasing the time‐efficiency of the entire procedure. Extensive tests were conducted in six different outdoor scenarios to show the performance of our prototype and method. To the best knowledge of the authors, this is the first time that such a design and method was presented and successfully tested on a legged robot.