We proposed a new methodology for learning and adaption of manipulation skills that involve physical contact with the environment. The proposed algorithm takes a reference Cartesian trajectory and force/torque profile as input and adapts the movement so that the resulting forces and torques match the reference profiles. The learning algorithm is based on dynamic movement primitives and quaternion representation of orientation, which provide a mathematical machinery for efficient and stable adaptation. Experimentally we showed that the robot’s performance can be significantly improved within a few iteration steps, compensating for vision and other errors that might arise during the execution of the task.
COBISS.SI-ID: 28552743
In this paper we presented a hierarchical framework for representation of manipulation actions and its applicability to the problem of top down action extraction from observation. The framework consists of novel probabilistic semantic models, which encode contact relations as probability distributions over the action phase. The models are action descriptive and can be used to provide probabilistic similarity scores for newly observed action sequences. The lower level of the representation consists of parametric hidden Markov models, which encode trajectory information.
COBISS.SI-ID: 28214567
We proposed an integrated policy learning framework that fuses iterative learning control (ILC) and reinforcement learning. Integration is accomplished at the exploration level of the reinforcement learning algorithm. The proposed algorithm combines fast convergence properties of iterative learning control and robustness of reinforcement learning. This way, the advantages of both approaches are retained while overcoming their respective limitations.
COBISS.SI-ID: 28650023
Study Objectives: Periodic breathing is sleep disordered breathing characterized by instability in the respiratory pattern that exhibits an oscillatory behavior. Periodic breathing is associated with increased mortality, and it is observed in a variety of situations, such as acute hypoxia, chronic heart failure, and damage to respiratory centers. The standard quantification for the diagnosis of sleep related breathing disorders is the apnea-hypopnea index (AHI), which measures the proportion of apneic/ hypopneic events during polysomnography. Determining the AHI is labor-intensive and requires the simultaneous recording of airflow and oxygen saturation. In this paper, we propose an automated, simple, and novel methodology for the detection and qualification of periodic breathing: the estimated amplitude modulation index (eAMI). Patients or Participants: Antarctic Cohort (3800 meters): 13 normal individuals. Sleep Clinic Cohort: 39 different patients suffering from diverse sleep-related pathologies. Measurements and Results: When tested in a population with high levels of periodic breathing (Antarctic Cohort), eAMI was closely correlated with AHI (r = 0.95, P ( 0.001). When tested in the clinical setting, the proposed method was able to detect portions of the signal in which subclinical periodic breathing was validated by an expert (n = 93; accuracy = 0.85). Average eAMI was also correlated with the loop gain for the combined clinical and Antarctica cohorts (r = 0.58, P ( 0.001). Conclusions: In terms of quantification and temporal resolution, the eAMI is able to estimate the strength of periodic breathing and the underlying loop gain at any given time within a record. The impaired prognosis associated with periodic breathing makes its automated detection and early diagnosis of clinical relevance.
COBISS.SI-ID: 2076588
The study evaluated the efficiency of two heat dissipation strategies under simulated desert patrol missions. Ten men participated in four trials, during which they walked on a treadmill (45°C, 20% relative humidity), carrying a load of 35 kg; two 50-min walks were separated by a 20-min rest. Cooling strategies, provided by an ambient air-ventilated vest (active cooling condition, AC), or water spraying of the skin during the rest (passive cooling condition, PC), in addition to reduced clothing and open zippers, were compared to conditions with full protective (FP) clothing and naked condition (NC). Skin temperature was higher during NC (37.9 +/- 0.4°C; p ( 0.001), and rectal temperature and heart rate were higher during FP (38.6 +/- 0.4°C, p ( 0.001 and 145 +/- 12, p ( 0.001, respectively), compared to other conditions. Four subjects terminated the trial prematurely due to signs of heat exhaustion in FP. Both cooling strategies substantially improved evaporative cooling. Practitioner Summary: Cooling strategies, provided by an ambient air-ventilated vest and water spraying of the skin, were compared to conditions with full protective clothing and a naked condition during a simulated desert patrol mission. Both cooling strategies improved evaporative cooling and reduced heat strain, compared to the full protection condition.
COBISS.SI-ID: 28850471