Darwin Caldwell - Selected Publications#

C Semini, N.G. Tsagarakis, E. Guglielmino, M.Focchi, F. Cannella and D.G. Caldwell, “Design of HyQ – a Hydraulically and Electrically Actuated Quadruped Robot”, Proc. IMechE.Vol.225 (6), Part I: Journal of Systems Engineering and Control, pp.831-849, (2011), (801 Citations). Describes the development of the HyQ quadruped robot, one of the most influential robots in the domain.

N. G. Tsagarakis, D. G. Caldwell, et al, “WALK-MAN: A High Performance Humanoid Platform for Realistic Environments”, Journal of Field Robotics 34(7), 1225–1259 (2017), DOI: 10.1002/rob.21702 (255 citations). Describes the development of the WalkMan robot that competed in the DARPA robotic challenge. This robot was the only EU funded (Walkman project) robot in the competition and was designed, built and operational in under 15 months.

C.Mastalli, I.Havoutis, M.Focchi, D.G.Caldwell, Claudio Semini , "Motion Planning for Quadrupedal Locomotion: Coupled Planning, Terrain Mapping and Whole-Body Control", IEEE Trans. on Robotics (T-RO): 36 (6), pp.1635-1648, (2020) DOI: 10.1109/TRO.2020.3003464. (104 Citations). Describes control software for the HyQ robot and particularly algorithms for Whole Body Locomotions. Humanoids and quadrupeds have many more than the 6 DOF found in industrial robots and solving for the complexity of these systems is both challenging and critical. These algorithms are vital to quadrupedal locomotion and this is demonstrated in this work.

P.Guadagna, M.Fernandes, F.Chen, D.G.Caldwell, S. Poni, C. Semini & M. Gatti, “Using deep learning for pruning region detection and plant organ segmentation in dormant spur-pruned grapevines”, Precision Agric. Volume 24, pages 1547–1569 (2023). DOI: 10.1007/s11119-023-10006-y. (9 citations) Describes the use of a legged robot (HyQ-Real) in a real world farming (vine pruning) task involving locomotion, vision, AI and manipulation. The shortage of workers for farming is a key societal challenge and robots have the potential to address this shortfall, providing food supply security.

D. Park, S.Toxiri, G.Chini, C. Di Natali, D.G. Caldwell, J.Ortiz, "Shoulder-sideWINDER (Shoulder-side Wearable INDustrial Ergonomic Robot): Design and Evaluation of Shoulder Wearable Robot with Mechanisms to compensate for Joint Misalignment", IEEE Transactions on Robotics (T-RO) (2022), Vol. 38(3), pp.1460-1471 DOI: 10.1109/TRO.2021.3125854. (19 Citations). Describes the design and testing in a real industrial setting of a new exoskeleton shoulder demonstrating high TRL capacity.

C. Di Natali, T. Poliero, M. Sposito, E. Graf, C. Bauer, C. Pauli, E. Bottenberg, A. De Eyto, L. O’Sullivan, A. F. Hidalgo, D. Scherly, K. S. Stadler, D. G. Caldwell, J. Ortiz, "Design and evaluation of a soft assistive lower limb exoskeleton", Robotica. Vol.37 (12), pp. 2014–2034, (2019). doi:10.1017/S0263574719000067 (91 Citations). Describes the design and operation of the XoSoft systems. This is one of the world first soft exoskeletons (Exosuits) and has been used as a basis for medical rehabilitation and may form a solution for muscle and bone loss in astronauts (working with ESA).

A.Acemoglu, J.Krieglstein, D.G. Caldwell, F.Mora, L.Guastini, M.Trimarchi, A.Vinciguerra, J.Hysenbelli, M.Delsanto, O.Barboni, S.Baggioni, G.Peretti, and L.S. Mattos, “5G Robotic Telesurgery – Proof of Concept Operation on a Cadaver Patient”, IEEE Transactions on Medical Robotics and Bionics, vol. 2, no. 4, pp. 511-518, Nov. (2020), DOI: 10.1109/TMRB.2020.3033007. (36 Citations). Describes the systems that performed the world's first 5G robotics surgery between central Milan and San Raffaele Hospital.

F.Larradet, R.Niewiadomski, G.Barresi, D.G.Caldwell and L.S.Mattos (2020), “Toward Emotion Recognition From Physiological Signals in the Wild: Approaching the Methodological Issues in Real-Life Data Collection”, Front. Psychol. 11, 1111, (2020), doi: 10.3389/fpsyg.2020.01111. (64 citations). Describes Real-world collection and assessment of emotions that can be applied in control and analysis of human-robot interactions. Typically data is collected in lab. Being able to collect real-world data will have a profound impact on the quality of data collected and available.

Y.Huang, L.Rozo, J.Silverio, and D.G.Caldwell,“Kernelized movement primitives” The International Journal of Robotics Research, vol. 38, no. 7, pp. 833–852, (2019). DOI:10.1177/0278364919846363 (169 Citations). Describes a novel approach to robot learning which allows the robot to adapt to learned human taught motor skills, and subsequently perform diverse tasks under varying physical and performance constraints. The kernelized motion primitive concept limits the data size of the learning frame making real-time operation feasible. This learning is more closely related to human learning experiences.

N. Jaquier, L. Rozo, D.G. Caldwell, S.Calinon,"Geometry-aware Manipulability Learning, Tracking and Transfer”, Int. Journal of Robotics Research (IJRR), 40 (2-3), pp.624-650, (2021) DOI: 10.1177/0278364920946815. (52 citations). This describes a novel manipulability transfer framework that allows robots (particularly bi-manual robots) to learn and reproduce manipulability ellipsoids from expert demonstrations. This has been applied and tested on humanoid and centaur robots such as COMAN, and Centauro and will have a fundamental impact on loco-manipulation capacity of high complexity (>10 dof) and ultra high complexity (>30 dof) robots.