Complex Task Performance Research Articles

An Analysis of Air Traffic Controllers’ Job Satisfaction

The air traffic controllers' job is one of the most hectic in today’s world, predominantly due to its safety-critical operations and altering expectations. The primary purpose of this paper is to provide a holistic directory of determinants and synthesized reinforcements for air traffic controllers' job satisfaction. Researchers in the past have put the spotlight on individual air traffic controller’s technical job satisfaction factors, such as impacts from remote tower operation, airplane trajectory changes, and dynamic air traffic situations. However, none described the connection among those factors and how adjusting those factors can enhance the cognitive components related to their job satisfaction. This paper's in-depth analysis identified factors contributing to air traffic controllers' job satisfaction based on past literature. It is intended to increase understanding and improve knowledge for future researchers and practitioners. The five predominant factors identified for air traffic controllers’ job satisfaction are ambiguity of job functions, overwhelming workload, complex task performance and uncertain work demand, job fatigue, and work-family conflict. Some effective methods to increase air traffic controllers’ job satisfaction are regular break between shifts, technological advancement to facilitate jobs, and sound insulations.

English learners' orientations towards complexity, accuracy, and fluency: Learner factors and L2 writing performance

Complexity, accuracy, and fluency (CAF) represent distinctive dimensions of language development, which have been widely used to assess learners’ task performance. Despite much attention to CAF constructs in research, little is known about how they are perceived by L2 learners. This study examines what CAF-specific orientations learners possess as they relate to their motivational self and anxiety, and it further explores how learners’ CAF orientations affect their L2 writing performance. Fifty-one Korean learners of English completed a questionnaire that gauged learners’ orientations towards CAF constructs. They also performed simple and complex writing tasks with a one week interval. The participants’ responses were submitted to factor analysis, and the writing performance was assessed in terms of CAF measures. The results revealed the convergence of learners’ CAF orientations, suggesting that the CAF triad exists as a unitary construct for L2 learners. Also, the participants’ CAF-orientation did not affect their writing performance in either simple or complex tasks.

Amyloid accumulation associated with worse performance of complex task

AbstractBackgroundThe preclinical stage of Alzheimer disease (AD) is largely considered a clinically silent period that is detectable primarily through neuroimaging and biofluid biomarkers of amyloid and tau. We evaluated whether performance of complex daily activities is associated with amyloid accumulation as measured by positron emission tomography (PET) among cognitively normal (CN) individuals with or without preclinical AD.MethodCN participants (Clinical Dementia Rating 0) from the Knight Alzheimer Disease Research Center performed three activity subtasks (shopping, checkbook balancing, and medication management) from the Performance Assessment of Self‐Care Skills (PASS) at their homes. They also had an amyloid PET scan using Pittsburgh Compound‐B within one year of completing the PASS. PASS subtasks were scored for independence, adequacy, and safety of performance. Nonparametric correlations were used to determine the associations between amyloid accumulation and performance on each subtask.ResultFifty‐six CN participants (mean age 74.8 years, 61% female, 87% white) were included. Median amyloid PET was 1.17 (interquartile range 1.07–1.63; standard cutoff for preclinical AD: 1.42). Amyloid PET accumulation was not associated with performance on the checkbook balancing or medication management subtasks of the PASS. Higher amyloid PET accumulation was associated with worse performance on the shopping subtask of the PASS.ConclusionThis study suggests that a complex task, measured using a performance‐based test, may be associated with higher amyloid deposition. Performance‐based measures could potentially serve as a less expensive method for screening individuals who may likely benefit from further evaluation (plasma amyloid or PET amyloid). They could also lead to a better understanding of functional changes that may be occurring during the preclinical stage of AD.

P041 The Impact of Simulated Night Shifts on Executive Functioning

Abstract Nightwork is associated with fatigue, decreased sleep quality, and impairments in cognitive function. While attentional tasks have been widely investigated, there are limited data on more complex tasks, such as executive functioning during nightwork. Workers often need to rapidly shift between tasks, adapting to new and complex situations. The aim of this study was to investigate the impact of nightwork on executive functioning. Healthy, non-shift working individuals (N=8; 5F, 24.8±5.0y) participated in a 7-day live-in laboratory study. Participants underwent an 8h TIB baseline sleep, followed by 4 consecutive simulated nightshifts with 7h TIB sleep during the day and an 8h TIB recovery sleep. Participants were assessed for executive function at 2000h, 2200h, 0100h and 0400h. Executive functioning was assessed with a mental flexibility switching task where a 3D rotation and math task were displayed simultaneously with an arrow indicating which task to complete in a random order. Resulting throughput data were analysed using linear mixed models. There was a main effect of time of night (F(3,77)=4.81,p=.004) on throughput such that there was a speed accuracy trade off over the night shift with slower switching ability later in the shift. There was also a main effect of nightshift (F(2,77)=54.33,p<.001) where participants’ performance improved on the task with each nightshift. This study suggests executive functioning is impaired on nightshift with worse performance at 0400h. Task improvements over consecutive nightshifts may have been due to learning or acclimation to nightwork. Understanding complex task performance on nightshift is important for tailoring countermeasures.

Elements of Conceptual Thinking in Planning the Behavior of Autonomous Intelligent Agents

The expediency of using the tools of visual-effective, visual-figurative and conceptual thinking for planning the purposeful activity of autonomous intelligent agents in problem environments of various degrees of a priori uncertainty has been substantiated. The content is revealed and the role of each form of thinking is shown in the process of automatic planning of the purposeful behavior of autonomous intelligent agents in the changing conditions of functioning. The special role of conceptual thinking in the performance of complex tasks by autonomous agents and the planning of polyphasic behavior associated with it is indicated. Taking into account the complexity of the problems associated with the formalization of mental acts of conceptual thinking, possible ways of its gradual development from the initial level to the transition to higher levels of development are shown, expanding on this basis the class of tasks solved by autonomous intelligent agents. A model of knowledge representation and tools for deriving solutions of the initial level of conceptual thinking have been developed, which allow intelligent agents to break down the tasks they receive into sub-goals of behavior. Then, on this basis, plan polyphase activity by searching for solutions to the associated subtasks, which ensure the determination of the minimum length routes of movement in a prob lematic environment with obstacles and the purposeful manipulation of objects in it. The tools are synthesized allowing to establish the order of elaboration of complex actions included in the structure of the task formulated by autonomous intelligent agents. It is shown that the further development of the proposed methodological foundations for constructing intelligent problem sol vers is associated with the formalization of a higher level of mental acts of conceptual thinking, which make it possible to solve practical problems of different complexity, formulated both in procedural and declarative form of presentation in the form of various target situations of the problem environment, having a large dimension.

Functional Connectivity Within the Fronto-Parietal Network Predicts Complex Task Performance: A fNIRS Study.

Performance in complex tasks is essential for many high risk operators. The achievement of such tasks is supported by high-level cognitive functions arguably involving functional activity and connectivity in a large ensemble of brain areas that form the fronto-parietal network. Here we aimed at determining whether the functional connectivity at rest within this network could predict performance in a complex task: the Space Fortress video game. Functional Near Infrared Spectroscopy (fNIRS) data from 32 participants were recorded during a Resting-State period, the completion of a simple version of Space Fortress (monotask) and the original version (multitask). The intrinsic functional connectivity within the fronto-parietal network (i.e., during the Resting-State) was a significant predictor of performance at Space Fortress multitask but not at its monotask version. The same pattern was observed for the functional connectivity during the task. Our overall results suggest that Resting-State functional connectivity within the fronto-parietal network could be used as an intrinsic brain marker for performance prediction of a complex task achievement, but not for simple task performance.

Voltage Control-Based Ancillary Service Using Deep Reinforcement Learning

Ancillary services rely on operating reserves to support an uninterrupted electricity supply that meets demand. One of the hidden reserves of the grid is in thermostatically controlled loads. To efficiently exploit these reserves, a new realization of control of voltage in the allowable range to follow the set power reference is proposed. The proposed approach is based on the deep reinforcement learning (RL) algorithm. Double DQN is utilized because of the proven state-of-the-art level of performance in complex control tasks, native handling of continuous environment state variables, and model-free application of the trained DDQN to the real grid. To evaluate the deep RL control performance, the proposed method was compared with a classic proportional control of the voltage change according to the power reference setup. The solution was validated in setups with a different number of thermostatically controlled loads (TCLs) in a feeder to show its generalization capabilities. In this article, the particularities of deep reinforcement learning application in the power system domain are discussed along with the results achieved by such an RL-powered demand response solution. The tuning of hyperparameters for the RL algorithm was performed to achieve the best performance of the double deep Q-network (DDQN) algorithm. In particular, the influence of a learning rate, a target network update step, network hidden layer size, batch size, and replay buffer size were assessed. The achieved performance is roughly two times better than the competing approach of optimal control selection within the considered time interval of the simulation. The decrease in deviation of the actual power consumption from the reference power profile is demonstrated. The benefit in costs is estimated for the presented voltage control-based ancillary service to show the potential impact.

The effects of observation in video games: how remote observation influences player experience, motivation, and behaviour

ABSTRACT Surveillance is ubiquitous. It is well known that the presence of other people (in-person or remote, actual or perceived) increases performance on simple tasks and decreases performance in complex tasks (Zajonc 1965). But little is known about these phenomena in the context of video games, with recent advances finding that they do not necessarily extend to games (Emmerich and Masuch 2018). In Experiment 1 (N=1489; No Observation vs. Researcher Observing), we find that participants observed by a researcher played significantly longer, and performed significantly better, across three video games. Moreover, we find some support that participants observed by a researcher score higher on player experience and intrinsic motivation. In Experiment 2 (N=843; Researcher Observing vs. Professor Observing), we seek to understand whether different roles differing in their perceived evaluativeness would influence the effects of observation. We find that participants observed by a professor had, at times, significantly lower performance, player experience, intrinsic motivation, playing time, and higher anxiety. In Experiment 3 (N=1358; No Observation vs. Researcher Observing), we further validate Experiment 1 by extending our results to three additional game genres. Here, we provide the largest study on observation in video games to date. The study is also the first to show that observer type can significantly influence player outcomes.

Research on Route Planning based on improved Ant Colony Algorithm

In order to enhance the safe flight performance of UAV and better complete complex tasks, it is extremely necessary to study route planning. Firstly, the feature points of the landform landform are extracted as the flight track points of the UAV, and the track planning problem is transformed into the traveling salesman problem. Secondly, an adaptive pheromone update method is proposed, and an improved ant colony algorithm for optimizing the local pheromone is proposed to avoid the basic ant colony algorithm from falling into the local optimal solution and improve the global search ability. Finally, the proposed algorithm is verified by simulation experiments. The experimental results show that the improved ant colony algorithm is practical and superior, and can effectively solve the problem of route planning and obtain the optimal path of route planning.

Performance in complex life situations: effects of age, cognition, and walking speed in virtual versus real life environments

BackgroundVirtual reality (VR) enables objective and accurate measurement of behavior in ecologically valid and safe environments, while controlling the delivery of stimuli and maintaining standardized measurement protocols. Despite this potential, studies that compare virtual and real-world performance of complex daily activities are scarce. This study aimed to compare cognitive strategies and gait characteristics of young and older healthy adults as they engaged in a complex task while navigating in a real shopping mall and a high-fidelity virtual replica of the mall.MethodsSeventeen older adults (mean (SD) age = 71.2 (5.6) years, 64% males) and 17 young adults (26.7 (3.7) years, 35% males) participated. In two separate sessions they performed the Multiple Errands Test (MET) in a real-world mall or the Virtual MET (VMET) in the virtual environment. The real-world environment was a small shopping area and the virtual environment was created within the CAREN™ (Computer Assisted Rehabilitation Environment) Integrated Reality System. The performance of the task was assessed using motor and physiological measures (gait parameters and heart rate), MET or VMET time and score, and navigation efficiency (cognitive performance and strategy). Between (age groups) and within (environment) differences were analyzed with ANOVA repeated measures.ResultsThere were no significant age effects for any of the gait parameters but there were significant environment effects such that both age groups walked faster (F(1,32) = 154.96, p < 0.0001) with higher step lengths (F(1,32) = 86.36, p < 0.0001), had lower spatial and temporal gait variability (F(1,32) = 95.71–36.06, p < 0.0001) and lower heart rate (F(1,32) = 13.40, p < 0.01) in the real-world. There were significant age effects for MET/VMET scores (F(1,32) = 19.77, p < 0.0001) and total time (F(1,32) = 11.74, p < 0.05) indicating better performance of the younger group, and a significant environment effect for navigation efficiency (F(1,32) = 7.6, p < 0.01) that was more efficient in the virtual environment.ConclusionsThis comprehensive, ecological approach in the measurement of performance during tasks reminiscent of complex life situations showed the strengths of using virtual environments in assessing cognitive aspects and limitations of assessing motor aspects of performance. Difficulties by older adults were apparent mainly in the cognitive aspects indicating a need to evaluate them during complex task performance.

Motor learning while performing a motor task measured by magnetic resonance imaging: a systematic review

Different variables, such as repetition and cognitive load, may explain the neurophysiological differences observed from one task to another in motor learning. This learning can be measured with functional magnetic resonance imaging. The aim of this systematic review was to document motor learning by functional magnetic resonance imaging during the performance of different simple or complex motor tasks in healthy subjects. The search for articles was carried out in the MEDLINE, PEDro, CINHAL and EBSCO databases in May 2020. The systematic review followed the PRISMA criteria. Nine studies were selected for a qualitative analysis. The quality of the studies ranged from 5 to 7 points on the PEDro scale. The qualitative analysis shows strong evidence that after repeating a motor task a motor learning process is generated. There is both strong and moderate evidence to show that action observation and sleep restriction are involved in motor learning. The results on sensory discrimination training were controversial. The results show, with high quality evidence, that repetition of a motor task is associated with the learning process, which seems to be related to a thickening of the motor cortex after the intervention measured with functional magnetic resonance imaging. These results are not conclusive, owing to the limiting factors of this systematic review.

Acute Alcohol Intake Deteriorates Hearing Thresholds and Speech Perception in Noise

Objectives: The hearing process involves complex peripheral and central auditory pathways and could be influenced by various situations or medications. To date, there is very little known about the effects of alcohol on the auditory performances. The purpose of the present study was to evaluate how acute alcohol administration affects various aspects of hearing performance in human subjects, from the auditory perceptive threshold to the speech-in-noise task, which is cognitively demanding. Methods: A total of 43 healthy volunteers were recruited, and each of the participants received calculated amounts of alcohol according to their body weight and sex with a targeted blood alcohol content level of 0.05% using the Widmark formula. Hearing was tested in alcohol-free conditions (no alcohol intake within the previous 24 h) and acute alcohol conditions. A test battery composed of pure-tone audiometry, speech reception threshold (SRT), word recognition score (WRS), distortion product otoacoustic emission (DPOAE), gaps-in-noise (GIN) test, and Korean matrix sentence test (testing speech perception in noise) was performed in the 2 conditions. Results: Acute alcohol intake elevated pure-tone hearing thresholds and SRT but did not affect WRS. Both otoacoustic emissions recorded with DPOAE and the temporal resolution measured with the GIN test were not influenced by alcohol intake. The hearing performance in a noisy environment in both easy (−2 dB signal-to-noise ratio [SNR]) and difficult (−8 dB SNR) conditions was decreased by alcohol. Conclusions: Acute alcohol elevated auditory perceptive thresholds and affected performance in complex and difficult auditory tasks rather than simple tasks.

An Energy-Efficient Silicon Photonic-Assisted Deep Learning Accelerator for Big Data

Deep learning has become the most mainstream technology in artificial intelligence (AI) because it can be comparable to human performance in complex tasks. However, in the era of big data, the ever-increasing data volume and model scale makes deep learning require mighty computing power and acceptable energy costs. For electrical chips, including most deep learning accelerators, transistor performance limitations make it challenging to meet computing’s energy efficiency requirements. Silicon photonic devices are expected to replace transistors and become the mainstream components in computing architecture due to their advantages, such as low energy consumption, large bandwidth, and high speed. Therefore, we propose a silicon photonic-assisted deep learning accelerator for big data. The accelerator uses microring resonators (MRs) to form a photonic multiplication array. It combines photonic-specific wavelength division multiplexing (WDM) technology to achieve multiple parallel calculations of input feature maps and convolution kernels at the speed of light, providing the promise of energy efficiency and calculation speed improvement. The proposed accelerator achieves at least a 75x improvement in computational efficiency compared to the traditional electrical design.

From the Savannah to the Corporate Office: The Evolution of Teams

Just as social and organizational research has benefited significantly from evolutionary approaches, likewise an understanding of the evolution of teams can advance our knowledge of team formation and functioning. The current paper traces the multilevel emergence of teams as a unique type of group that evolved specially for complex task performance, outlines the evolved mechanisms that enable humans to use teams as a form of adaptive technology, and describes how teams leverage flexible structural adjustments and distinct human motives to tackle a wide range of challenges. We conclude by discussing the implications of this novel framework for team research.

Goal-Oriented Attention Self-Regulation Training Improves Executive Functioning in Veterans with Post-Traumatic Stress Disorder and Mild Traumatic Brain Injury.

Difficulties in executive-control functions are common sequelae of both traumatic brain injury (TBI) and post-traumatic stress disorder (PTSD). The goal of this study was to assess whether a cognitive rehabilitation training that was applied successfully in civilian and military TBI would be effective for military Veterans with comorbid PTSD and mild TBI (mTBI). In the previous study, Veterans with a history of mild to severe TBI improved significantly after goal-oriented attentional self-regulation (GOALS) training on measures of attention/executive function, functional task performance, and emotional regulation. The objective of this study was to assess effects of GOALS training in Veterans with comorbid PTSD and mTBI. Forty Veterans with a current PTSD diagnosis and history of mTBI (6+ months post) were randomized to either five weeks of GOALS or Brain-Health Education (BHE) training matched in time and intensity. Evaluator-blinded assessments at baseline and post-training included neuropsychological and complex functional task performance, and self-report measures of emotional functioning/regulation. After GOALS but not BHE training, participants significantly improved from baseline on primary outcome measures of: overall complex attention/executive function neuropsychological performance composite (F = 12.35, p = 0.001; Cohen d = 0.48), and overall mood disturbance -POMS emotional regulation self-report (F = 4.29, p = 0.05, Cohen d = 0.41). In addition, GOALS but not BHE participants indicated a significant decrease in PTSD symptoms (PCL-M Total Score) (F = 4.80, p = 0.05, Cohen d = 0.60), and demonstrated improvement on complex functional task performance-GPS Learning and Memory (F = 5.06, p = 0.05, Cohen d = 0.56]. Training in attentional self-regulation applied to participant-defined goals may improve cognitive functioning in Veterans with comorbid PTSD and mTBI. Improving cognitive control functioning may also improve functioning in other domains such as emotional regulation and functional performance, potentially making it particularly relevant for Veterans with a history of mTBI and comorbid psychiatric symptoms.

Modularity in Motor Control: Similarities in Kinematic Synergies Across Varying Locomotion Tasks.

Kinematic synergies (kSYN) provide an approach to quantify the covariation of joint motions and to explain the mechanisms underlying human motor behavior. A low-dimensional control strategy by means of the activation of a moderate number of kSYN would simplify the performance of complex motor tasks. The purpose of this study was to examine similarities between the kSYN of varying locomotion tasks: straight-line walking, walking a 90° spin turn and walking upstairs. Task-specific kSYN were extracted from full body kinematic recordings of 13 participants by principal component analysis. The first five kSYN accounting for most of the variance within each task were selected for further analysis following previous studies. The similarities between the kSYN of the three different locomotion tasks were quantified by calculating cosine similarities (SIM), as a vector-based similarity measure ranging from 0 (no similarity) to 1 (high similarity), between absolute principal component loading vectors. A SIM between two kSYN > 0.8 was interpreted as highly similar. Two to three highly similar kSYN were identified when comparing two individual tasks with each other. One kSYN, primarily characterized by anteversion and retroversion of the arms and legs, were found to be similar in all three tasks. Additional kSYN that occurred between individual tasks reflected mainly an upwards/downwards movement of the body or a countercyclical knee flexion/extension. The results demonstrate that the three investigated locomotion tasks are characterized by kSYN and that certain kSYN repeatedly occur across the three locomotion tasks. PCA yields kSYN which are in descent order according to their amount of total variance accounted for. Referring to the placing of a kSYN within the order as priorization, we found a change in priorization of repeatedly occurring kSYN across the individual tasks. The findings support the idea that movements can be efficiently performed through a flexible combination of a lower number of control-relevant variables.

Symptoms of convergence and accommodative insufficiency predict engagement and cognitive fatigue during complex task performance with and without automation

Deficits in the accommodative and/or vergence responses have been linked with inattentive behavioral symptoms. While using automated systems (e.g., self-driving cars, autopilot), operators (e.g., drivers, pilots, soldiers) visually monitor displays for critical changes, making deficits in the accommodative and/or vergence responses potentially hazardous for individuals remaining actively engaged in the task at hand. The purpose of this study was to determine if symptoms of accommodative-vergence deficits predict an individual's level of task engagement and cognitive fatigue while performing a flight simulation task with or without automation. Eighty-four participants performed a flight simulation task with or without automation. Prior to task completion, self-report accommodative-convergence deficit symptoms were assessed with the Convergence Insufficiency Symptom Survey (CISS). Before and after the flight simulation task participants rated their task engagement and cognitive fatigue. Electroencephalographic activity (EEG) was recorded concurrently during task performance. Results showed that higher scores on the CISS were related to increased feelings of fatigue and decreased ratings of task engagement. The CISS was also positively related to parietal-occipital fast alpha power during the last 10 min of the task for participants using automation, suggesting increased cortical idling. CISS scores did not predict performance. Results have implications for optimizing operator cognitive states over extended task performance.

Degeneracy and Complexity in Neuro-Behavioral Correlates of Team Coordination.

Team coordination—members of a group acting together rather than performing specific actions individually—is essential for success in many real-world tasks such as military missions, sports, workplace, or school interactions. However, team coordination is highly variable, which is one reason why its underlying neural processes are largely unknown. Here we used dual electroencephalography (EEG) in dyads to study the neurobehavioral dynamics of team coordination in an ecologically valid task that places intensive demands on joint performance. We present a novel conceptual framework to interpret neurobehavioral variability in terms of degeneracy, a fundamental property of complex biological systems said to enhance flexibility and robustness. We characterize degeneracy conceptually in terms of a manifold representing the geometric locus of the dynamics in the high dimensional state-space of neurobehavioral signals. The geometry and dimensionality of the manifold are determined by task constraints and team coordination requirements which restrict the manifold to trajectories that are conducive to successful task performance. Our results indicate that team coordination is associated with dimensionality reduction of the manifold as evident in increased inter-brain phase coherence of beta and gamma rhythms during critical phases of task performance where subjects exchange information. Team coordination was also found to affect the shape of the manifold manifested as a symmetry breaking of centro-parietal wavelet power patterns across subjects in trials with high team coordination. These results open a conceptual and empirical path to identifying the mechanisms underlying team performance in complex tasks.

High-Level Motor Planning Assessment During Performance of Complex Action Sequences in Humans and a Humanoid Robot

Examining complex cognitive-motor performance in humanoid robots and humans can inform their interactions in a social context of team dynamics. Namely, the understanding of human cognitive-motor control and learning mechanisms can inform human motor behavior and also the development of intelligent controllers for robots when interacting with people. While prior humans and humanoid robot studies mainly examined motion planning, only a few have investigated high-level motor planning underlying action sequences for complex task execution. This sparse work has largely considered well-constrained problems using fairly simple performance assessment methods without detailed action sequence analyses. Here we qualitatively and quantitatively assess action sequences generated by humans and a humanoid robot during execution of two tasks providing various challenge levels and learning paradigms while offering flexible success criteria. The Levenshtein distance and its operators are adapted to the motor domain to provide a detailed performance assessment of action sequences by comparing them to a reference sequence (perfect sequence having a minimal number of actions). The results reveal that (i) humans produced a large variety of action sequences combining perfect and imperfect sequences while still reaching the task goal, whereas the robot generated perfect/near-perfect successful action sequences; (ii) the Levenshtein distance and the number of insertions provide reliable performance markers capable of differentiating perfect and imperfect sequences; (iii) the deletion operator is the most sensitive marker of action sequence failure. This work complements prior efforts for complex task performance assessment in humans and humanoid robots and has the potential to inform human–machine interactions.

HyP-DESPOT: A hybrid parallel algorithm for online planning under uncertainty

Robust planning under uncertainty is critical for robots in uncertain, dynamic environments, but incurs high computational cost. State-of-the-art online search algorithms, such as DESPOT, have vastly improved the computational efficiency of planning under uncertainty and made it a valuable tool for robotics in practice. This work takes one step further by leveraging both CPU and GPU parallelization in order to achieve real-time online planning performance for complex tasks with large state, action, and observation spaces. Specifically, Hybrid Parallel DESPOT (HyP-DESPOT) is a massively parallel online planning algorithm that integrates CPU and GPU parallelism in a multi-level scheme. It performs parallel DESPOT tree search by simultaneously traversing multiple independent paths using multi-core CPUs; it performs parallel Monte Carlo simulations at the leaf nodes of the search tree using GPUs. HyP-DESPOT provably converges in finite time under moderate conditions and guarantees near-optimality of the solution. Experimental results show that HyP-DESPOT speeds up online planning by up to a factor of several hundred in several challenging robotic tasks in simulation, compared with the original DESPOT algorithm. It also exhibits real-time performance on a robot vehicle navigating among many pedestrians.