Total Task Completion Research Articles

Conflict-free coordination planning for multiple automated guided vehicles in an intelligent warehousing system

With the rapid development of intelligent manufacturing, the application of automated guided vehicles (AGVs) in intelligent warehousing systems has become increasingly common. Efficiently planning the conflict-free paths of multiple AGVs while minimizing the total task completion time is crucial for evaluating the performance of the system. Distinguishing itself from recent approaches where conflict avoidance strategy and path planning algorithm are executed independently or separately, this paper proposes an improved conflict-free A* algorithm by integrating the conflict avoidance strategy into the initial path planning process. Based on the heuristic A* algorithm, we use the instruction time consumption as the key evaluation indicator of the cost function and add the turning consumption in the future path cost evaluation. Moreover, the expansion mode of child nodes is optimized where a five-element search set containing the “zero movement” is proposed to implement a proactive pause-wait strategy. Then the prediction rules are designed to add constraints to three types of instructions based on the timeline map, guiding the heuristic planning to search for conflict-free child nodes. Extensive simulations show that the coordination planning based on the improved conflict-free A* algorithm not only effectively achieves advanced conflict avoidance at the algorithmic level, but also exhibits lower computational complexity and higher task completion efficiency compared to other coordination planning methods.

Speed of performing complex movement tasks under decision-making conditions as a determinant of the tactical preparation level in kickboxers

This study aimed to evaluate the relationship between the speed of performing a complex motor task carried out under conditions requiring decision-making and the sports level of the kickboxers being studied. The authors constructed a test involving a complex author’s kickboxing task that mirrored the competitive conditions during a sports fight. Forty-seven K-1 kickboxing Polish National Team members (37 men and 10 women) performed a specific series of strikes across three variants. The relationship between the speed of these complex movement tasks, performed under decision-making conditions, and the sports level of the kickboxers, was evaluated. Deciding to start or change an action in reaction to external stimuli significantly (p < 0.001) increased the total task completion time in the male and female subject groups. The time spent deciding to take action and the time spent on deciding on the action were not significantly different. Quick execution of complex tasks under decision-making conditions, such as selecting the appropriate technical and tactical action, can become a decisive factor in determining a sports result. Making decisions to start or change an action as a reaction to external stimuli significantly (p < 0.001) extended the total time of task execution in the studied group. However, the time spent deciding to start the action and the time devoted to deciding during the action did not differ significantly. Therefore, the tactical solution an athlete uses makes no difference, though they must perform it confidently and with minimal time loss.

DEVELOPMENT AND ACCEPTABILITY OF A MOBILE APP FOR MONITORING EVERYDAY FUNCTIONAL ABILITIES

Abstract Subtle changes in instrumental activities of daily living (IADL), also termed everyday functional abilities, may be early markers of dementia. However, there are no widely available tools for tracking IADL function, and it is unclear whether such a tool would be used by older adults. The present study investigates older adults’ adherence to a novel mobile app designed to conveniently and cost-effectively monitor IADL function. We also examine participants’ acceptability of the app, including their overall satisfaction, perceived ease of use, benefits and limitations, and suggestions for improvement. Community-dwelling older adults (n=24; Mage=71.63 years) were randomly assigned three different simulated IADL tasks, to complete at specified times, each day for two weeks. Adherence was assessed using app usage data. App perceptions were assessed via survey and individual interview at the end of the two-week study period. Results indicate high adherence, with an average total task completion of 94.2%. Results from follow-up interviews revealed the app was generally perceived as “easy to use” and the assigned tasks could be completed quickly. Participants expressed frustration over app glitches that prevented them from successfully completing tasks and noted the assigned tasks and associated steps to complete were too simplified to be realistic. High adherence rates and general acceptance indicate the app’s potential usefulness. Further research should assess long-term effectiveness and user experience, especially among a more diverse sample. If reliable and valid, such apps could enable early dementia detection, but only if their use is feasible and accepted.

Design, Development, and Evaluation of an mHealth App for Reporting of Side Effects During Cytostatic Treatment: Usability Test and Interview Study.

Using mobile health (mHealth) interventions such as smartphone apps to deliver health services is an opportunity to engage patients more actively in their own treatment. Usability tests allow for the evaluation of a service by testing it out on the relevant users before implementation in clinical practice. The objective of this study was to design, develop, and evaluate the user interface of an app that would aid patients with cancer in reporting a more comprehensive summary of their side effects. The usability test was conducted by exposing patients with cancer to a prototype of an mHealth app that allowed for reporting of side effects from a chemotherapy regimen. After solving a set of 13 tasks, the test participants completed a system usability scale questionnaire and were interviewed using a semistructured interview guide. The interviews were later transcribed and analyzed. The 10 test participants had a mean age of 56.5 (SD 7.11) years. The mean total task completion time for the task-solving session was 240.15 (SD 166.78) seconds. The calculated system usability scale score was 92.5. Most participants solved most of the tasks without any major issues. A minority reported having difficulties using apps on smartphones in general. One patient never achieved a meaningful interaction with our app prototype. Most of those who engaged with the app approved of features that calmed them down, made them more empowered, and put them in control. They preferred to report on side effects in a detailed and concise manner. App features that provided specific advice could provoke both fear and rational action. The user tests uncovered design flaws that allowed for subsequent refining of an app that has the potential to enhance the safety of patients undergoing home-based chemotherapy. However, a refined version of the app is unlikely to be of value to all patients. Some might not be able to use apps on smartphones in general, or their ability to use apps is impaired because of their disease. This finding should have implications for health care providers' overall design of their follow-up service as the service must allow for all the patients to receive safe treatment whether they can use an mHealth app or not.

Joint Offloading and Resource Allocation With Diverse Battery Level Consideration in MEC System

Most of existing works related to mobile edge computing (MEC) focused on the total energy consumption or task completion time. However, for the applications with multiple players as a whole, the diverse battery levels of players should be considered to achieve better quality of experience (QoE), which is reflected in the metrics of task completion time and the lifespan of service. In this paper, we focus on QoE and diversity of devices. By jointly considering the individual battery level of different devices, we formulate an optimization problem to minimize the task completion time of the system. Firstly, we propose two necessary conditions to achieve the optimal solution of the formulated problem. In addition, we transform the formulated problem into another problem, which can be used to judge whether a given value is feasible for the formulated problem based on the proposed optimal conditions and bisection searching (BSS) algorithm. Then, we propose a method named sub-channel (SC) matching based on virtual computation resource (SMVCR) to solve the transformed problem with low computational complexity. Simulation results indicate that the BSS-SMVCR algorithm has a good performance in task completion time compared with some existing algorithms, and also can significantly extend the lifespan of the corresponding service, which is reflected in the supporting number of offloading tasks.

PASTO: Enabling Secure and Efficient Task Offloading in TrustZone-Enabled Edge Clouds

Offloading tasks to edge servers has been regarded as a potential way to solve the computation resource poverty problem of the end devices like autonomous vehicles. However, due to the sharing and openness features of edge computing, it raises severe security problems. In order to combat such issues, Trust Execution Environment (TEE), such as TrustZone, is advocated to strength the security of task offloading to edge computing in the hardware level. For the exploration of TrustZone, the inevitable involvement of cryptographic operations make existing offloading strategies not applicable, or not efficient enough, any more. In addition, TrustZone does not allow multiple tasks to coexist and execute on one CPU core at the same time. Taking the above issues into consideration, we investigate a secure task offloading problem for minimizing the total task completion time under the constraint of energy budget. To address this problem, we propose a Priority-aware Secure Task Offloading (PASTO) algorithm and evaluate the performance of PASTO by both numerical analysis and prototype based experiments. All the experiment results show that PASTO can effectively reduce the total task completion time in comparison with other state-of-the-art offloading approaches in TrustZone-enabled edge clouds.

Layer Dependency-Aware Learning Scheduling Algorithms for Containers in Mobile Edge Computing

Due to the features of lightweight and easy deployment, the use of containers has emerged as a promising approach for Mobile Edge Computing (MEC). Before running the container, an image composed of several layers must exist locally. However, it has been conspicuously neglected by existing work that task scheduling at the granularity of the layer instead of the image can significantly reduce the task completion time to further meet the real-time requirement and resource efficiency in resource-limited MEC. To bridge the gap, considering the complex dependency between layers and images, a novel layer dependency-aware container scheduling algorithm is proposed to reduce the total task completion time. Specifically: 1) We model the online layer dependency-aware scheduling problem for containers in a heterogeneous MEC, considering the layer download time and task computation time. 2) A policy gradient algorithm is proposed to solve this problem, and the high-dimensional and low-dimensional relations for layer dependencies are extracted with improved action selection. 3) Experiments based on the real-world data trace show that the proposed algorithm outperforms the image-based and layer-based baseline algorithms by 54% and 19% on average, respectively.

Hierarchical Deep Reinforcement Learning for Self-Powered Monitoring and Communication Integrated System in High-Speed Railway Networks

To align with the vision of future intelligent high-speed railway (HSR) networks, integrating sensor monitoring and remote communication are challenging for ensuring the lightweight of train equipment, high-quality transmissions, and dynamic interaction between monitoring and communication. In this paper, we propose a self-powered multisensor monitoring and communication integrated system in HSR. A low-power backscatter communication working framework of the self-powered monitoring system is designed in the monitoring network model, and a finite Gaussian mixture model (GMM) clustering method is used to analyze the communication cell coverage area in the communication network model. Aiming to minimize the total task completion time, we formulate a data monitoring and remote communication problem with the energy transfer constraint, data collection constraint, and transmission data rate constraint. As for the non-convex minimum time optimization problem, we develop a novel option-based hierarchical deep reinforcement learning (OHDRL) method to deal with the complex continuous variation characteristics of the monitoring and communication integrated HSR system. The system learns to select options at a high level, and the action is executed according to the policy of the selected option at a low level. This approach enables us to handle stochastic HSR environments, closed-loop policies, and goals in a temporal abstraction way. Numerical results reveal that the proposed algorithm for the integrated monitoring and communication HSR achieves a significantly higher reward and more stable learning performance than other algorithms in the literature.

A Step Towards Conditional Autonomy - Robotic Appendectomy

In recent years, Robot-Assisted Minimally Invasive Surgery (RAMIS) has been widely adopted worldwide due to its high precision, improved ergonomics and intuitive control. With the advances in artificial intelligence and surgical robot technologies, it is anticipated that the cognitive load on the surgeons could be relieved with a higher level of autonomy on surgical robots as well as simultaneously improving the precision of surgical robots' operations, minimising the risk of human errors and leading to safer operations. However, much research is still focused on task autonomy due to the complexity of the operations, and the limited clinical data on robotic surgery in some specific scenarios. Therefore, recent research has widely used simulated surgery scenarios to collect data to train artificial intelligence (AI) models. In this paper, we propose a method as a step towards conditional autonomy for robotic appendectomy. To achieve conditional autonomy, demonstration data was collected from human operators performing the appendectomy in a simulated robotic scene. The robotic instrument movements and trajectories for performing the appendectomy were learned from the demonstrated data. Once the surgeon specifies the location of the appendix, the proposed framework can carry out the appendectomy autonomously by using the dynamic motion primitives (DMP) generated from the learned surgical movements. Through the extensive validations in a simulated environment and on the da Vinci Research Kit (dVRK), we have shown that the proposed method can carry out the appendectomy procedure semi-automatically. Based on five metrics, it shows that the framework can reduce the total task path length, completion time and appendix stump length while maintaining a high degree of similarity to the demonstrated trajectory. This work demonstrated the feasibility of conditional autonomy in robotic surgeries.

Force-based assessment of tissue handling skills in simulation training for robot-assisted surgery

IntroductionAlthough robotic-assisted surgery is increasingly performed, objective assessment of technical skills is lacking. The aim of this study is to provide validity evidence for objective assessment of technical skills for robotic-assisted surgery.MethodsAn international multicenter study was conducted with participants from the academic hospitals Heidelberg University Hospital (Germany, Heidelberg) and the Amsterdam University Medical Centers (The Netherlands, Amsterdam). Trainees with distinctly different levels of robotic surgery experience were divided into three groups (novice, intermediate, expert) and enrolled in a training curriculum. Each trainee performed six trials of a standardized suturing task using the da Vinci Surgical System. Using the ForceSense system, five force-based parameters were analyzed, for objective assessment of tissue handling skills. Mann–Whitney U test and linear regression were used to analyze performance differences and the Wilcoxon signed-rank test to analyze skills progression.ResultsA total of 360 trials, performed by 60 participants, were analyzed. Significant differences between the novices, intermediates and experts were observed regarding the total completion time (41 s vs 29 s vs 22 s p = 0.003), mean non zero force (29 N vs 33 N vs 19 N p = 0.032), maximum impulse (40 Ns vs 31 Ns vs 20 Ns p = 0.001) and force volume (38 N3 vs 32 N3 vs 22 N3p = 0.018). Furthermore, the experts showed better results in mean non-zero force (22 N vs 13 N p = 0.015), maximum impulse (24 Ns vs 17 Ns p = 0.043) and force volume (25 N3 vs 16 N3p = 0.025) compared to the intermediates (p ≤ 0.05). Lastly, learning curve improvement was observed for the total task completion time, mean non-zero force, maximum impulse and force volume (p ≤ 0.05).ConclusionConstruct validity for force-based assessment of tissue handling skills in robot-assisted surgery is established. It is advised to incorporate objective assessment and feedback in robot-assisted surgery training programs to determine technical proficiency and, potentially, to prevent tissue trauma.

MARL based resource allocation scheme leveraging vehicular cloudlet in automotive-industry 5.0

Automotive-Industry 5.0 will use Beyond Fifth-Generation (B5G) communications to provide robust, abundant computation resources and energy-efficient data sharing among various Intelligent Transportation System (ITS) entities. Based on the vehicle communication network, the Internet of Vehicles (IoV) is created, where vehicles’ resources, including processing, storage, sensing, and communication units, can be leveraged to construct Vehicular Cloudlet (VC) to realize resource sharing. As Connected and Autonomous Vehicles (CAV) onboard computing is becoming more potent, VC resources (comprising stationary and moving vehicles’ idle resources) seems a promising solution to tackle the incessant computing requirements of vehicles. Furthermore, such spare computing resources can significantly reduce task requests’ delay and transmission costs. In order to maximize the utility of task requests in the system under the maximum time constraint, this paper proposes a Secondary Resource Allocation (SRA) mechanism based on a dual time scale. The request service process is regarded as M/M/1 queuing model and considers each task request in the same time slot as an agent. A Partially Observable Markov Decision Process (POMDP) is constructed and combined with the Multi-Agent Reinforcement Learning (MARL) algorithm known as QMix, which exploits the overall vehicle state and queue state to reach effective computing resource allocation decisions. There are two main performance metrics: the system’s total utility and task completion rate. Simulation results reveal that the task completion rate is increased by 13%. Furthermore, compared with the deep deterministic policy optimization method, our proposed algorithm can improve the overall utility value by 70% and the task completion rate by 6%.

A study on deep reinforcement learning-based crane scheduling model for uncertainty tasks

Abstract Aiming at the crane scheduling problem for uncertainty tasks in multi-crane scheduling situation, this article proposes a deep reinforcement learning-based crane scheduling modeling method that is not dependent on mathematical planning and has certain generality. First, the crane scheduling process is integrated into deep reinforcement learning framework in which the orbit space of the crane and the transportation task is environmental information and crane is the intelligent agent. Second, the interactive mode between reinforcement learning algorithm and environment is adjusted to adapt to the combined learning of multi-crane scheduling model. Last, the crane scheduling model for uncertainty tasks is constructed by optimizing reward discount factor, learning rate, and reward function intensive mode. Testing of the model is carried out based on practical crane scheduling in one steelmaking workshop. Scheduling proposal is generated and all crane tasks are completed within the planned time, which verifies the feasibility of this model. Results show that compared with manual scheduling plan, the scheduling proposal based on the new model reduces total task completion time by 11.52%, time of collision of crane routes decreases by 57.14%, and negative crane transportation distance shortens by 55.26%. The high efficiency of the scheduling model is therefore verified.

Reconnaissance and Confirmation Task Planning of Multiple Fixed-Wing UAVs with Specific Payloads: A Comparison Study

In this study, the reconnaissance and confirmation task planning of multiple fixed-wing unmanned aerial vehicles (UAV) with specific payloads, which is an NP-hard problem with strong constraints and mixed variables, is decomposed into two subproblems, task allocation with “payload-target” matching constraints, and fast path planning of the UAV group, for which two mathematical models are respectively established. A bi-layer collaborative solution framework is also proposed. The outer layer optimizes the allocation scheme between the UAVs and targets, whereas the inner layer generates the UAV path and evaluates the outer scheme. In the outer layer, a unified encoding based on the grouping and pairing relationship between UAVs and targets is proposed. The corresponding combinatorial mutation operators are then designed for the representative NSGA-II, MOEA/D-AWA, and DMOEA-ϵC algorithms. In the inner layer, an efficient heuristic algorithm is used to solve the path planning of each UAV group. The simulation results verify the effectiveness of the cooperative bi-layer solution scheme and the combined mutation operators. At the same time, compared with the NSGA-II and MOEA/D-AWA, DMOEA-ϵC can obtain a significantly better Pareto front and can weigh the assigned number of UAVs and the total task completion time to generate more diversified reconnaissance confirmation execution schemes.

NOMA-Enabled Computation and Communication Resource Trading for a Multi-User MEC System

In this article, we establish a novel framework for a multi-user mobile edge computing (MEC) network in which a set of users with high downlink rate demands and a set of users with intensive computation tasks can collaborate to achieve a mutually-beneficial scenario such that completion time of the tasks is reduced and the base station (BS) can send more information at a higher rate to the downlink users. Specifically, by leveraging non-orthogonal multiple access (NOMA) for uplink and downlink traffic, the user with the computation task can offload shares of the computation task to the edge cloud and the downlink user. At the same time, this user forwards the information it receives from the BS to the downlink user. In this set up, we jointly optimize the communication resources, computational resources at the edge cloud and user devices, pairings among the two sets of users, the shares of computation tasks, and relay bits to minimize the total task completion time while satisfying downlink users’ incentive requirements. For a network with a single computation demanding user and a single downlink user, the optimal solution to the problem is provided. For a network with multiple users, the problem is non-convex and computationally challenging. Hence, we propose an efficient, low complexity algorithm that utilizes the bottleneck matching algorithm, convex optimization, and the block coordinate descent scheme to obtain a locally-optimal solution. Simulation results demonstrate that, as compared with the state-of-the-art, the total task completion time is greatly reduced (32%–51%), and a large computational energy savings at the edge cloud (38%–55%) is achieved. Simultaneously, the downlink users’ rates improve compared to the orthogonal transmission.

A decision model on human-robot collaborative routing for automatic logistics

This paper studies an integrated human-robot routing problem arising in collaborative order picking systems. Travelling routes of pickers and robots interact with each other and simultaneously determine the start time of tasks. A mixed-integer programming model is proposed which aims to minimize the total task completion time by optimizing both pickers’ and robots’ routes. Moreover, a variable neighborhood search and dynamic programming -based heuristic algorithm is developed for solving the problem in large-scale environments. Numerical experiments are conducted to prove the necessity of the proposed model and also validate the efficiency of the proposed heuristic algorithm. The results indicate the methodology can yield solutions with smaller optimality gap within a reasonable period of time, in comparison with an intuitive but common decision rule. A sensitivity analysis is also conducted to derive a number of managerial implications regarding the configuration of pickers and robots and robots’ speed setting.

The Puzzles Test and the Red Shapes Test as new diagnostic tools for neglect syndrome

Introduction Neuropsychological methods for diagnosing neglect syndrome (NS) are focused on identifying the inability of patients to respond to stimuli localized in contralesional space. There are a large number of methods capable of diagnosing spatial neglect, but at the same time having various limitations and restrictions in their use.ObjectivesTo devise and to test universal diagnostic techniques for visuospatial neglect detection.Methods1) A.R. Luria test battery; Trail Making Test (Part A); the Bells Test; 2) Authors’ methods: the Puzzles Test, the Red Shapes Test. A total of 47 patients after stroke with right hemisphere damage participated in the study and were divided into a target (18 patients with NS) and a control (29 patients without NS) groups. The Puzzles Test consists of three tasks: turning over cards, completing a sentence using cards with letters, completing a picture. The Red Shapes Test consisted in the search for a variable number of geometric shapes. Objective indicators of the study: total task completion time, the number of left omissions.ResultsThe sensitivity of the tests to NS was examined using the Mann-Whitney U-test. Differences in the number of omissions and task completion time between patients with and without spatial neglect were statistically significant regarding all tasks: turning over cards (p=0.01), completing a sentence (p&lt;0.001), completing a picture (p&lt;0.001), finding geometric shapes (p&lt;0.01).ConclusionsThe Puzzles Test and Red Shapes Test along with the foreign tests (the Bells Test, Trail Making Test) are sufficiently effective methods for spatial neglect detection.DisclosureNo significant relationships.

Characterising users’ task completion process in learning-related tasks:A search pace model

This study investigated users’ searching, reading and writing interactions and their activity transitions during task completion process when users were collecting information for learning-related search tasks. Task completion process was defined as the process users started to search till the time when they have collected enough information to accomplish the search tasks. The data analysis was conducted from a new process perspective through synthesising macro- and micro-process levels. Four evenly distributed stages were divided according to the total task completion time in each search session. Our results demonstrated that users generally experienced three sub-processes during task completion process: exploration, accumulation and composition/reporting. Exploration sub-process is basically the first quarter of the total task completion time, during which users often issue more queries and visit more search engine result pages (SERPs) to collect information, and the dominant activity transition is switching between searching and reading; accumulation sub-process is mainly the second and third quarters of the total task completion time, during which they visit more unique content pages, have more revisits per content page, and they switch between reading and writing activities frequently; the last stage is composition/reporting sub-process, which is dominated by writing, and users often switch between writing and reading, and between writing and searching. Based on these findings, we propose a search pace model to describe how users proceed from the beginning to the end of task completion process in these three sub-processes. The methodology applied has been proved to be effective to examine users’ interaction behaviours from the process perspective on both the micro- and macro-levels. The findings of this article help us understand how users proceed their dynamic searching, reading and writing behaviours for learning-related tasks, and also have implications for the design of search systems that support learning-related tasks.

Development of a generic version of the multiple errands test for severe acquired brain injuries

Background Impairment of executive functions is a frequent and disabling consequence of brain injury. In 1991, Shallice and Burgess elaborated the Multiple Errand Test (MET) to assess everyday troubles associated with this impairment. Objective Our first aim was to create a generic version of MET (MET-G) readily adaptable to different settings and for retesting. The second aim was to validate the revised version to detect executive impairment in individuals with severe ABI. Methods An inter-professional team of experts in neurorehabilitation created a revised version of MET. Twenty-one patients with sABI (severe Acquired Brain Injury) and a control group of 20 neurologically healthy subjects, took part in the testing of MET-G. Results MET-G clearly distinguished people with sABI from healthy controls based on Total error score, task completion, and rule breaking. The test showed good inter-rater reliability and internal consistency. Conclusions The new, generic version of MET was able to differentiate adults with sABI from controls and proved to be a good tool for evaluating executive functions in these patients in daily-life contexts. Indications on how to adapt the test to different contexts and different scoring modalities are provided.

One- versus two-surgeon active stone retrieval procedures for flexible ureteroscopy: An off-site simulator comparative study.

To evaluate the efficacy of the one-surgeon basketing technique compared to the conventional two-surgeon method for flexible ureteroscopy. Seven urologists (three trained and four less-trained surgeons) extracted three renal stones using each technique with an off-site simulator. We analyzed the task completion time, the number of times the basket forceps were opened and closed, and unexpected events during stone-catching, for both techniques. There was no significant difference in the total task completion time between the one-surgeon (102.0 [interquartile range 63.5-164.0] s) and two-surgeon (99.5[interquartile range 75.0-145.3] s) techniques. However, the time it took the trained surgeons to extract all stones was significantly shorter with the one-surgeon than with the two-surgeon technique (19.0 [interquartile range 18.0-20.0]s vs 34.0 [interquartile range 25.0-40.0] s; P=0.049). Among the less-trained surgeons, the opposite effect was observed, but this difference was not statistically significant (96.5 [interquartile range 61.0-134.5] s vs 64.0 [interquartile range 51.5-77.0]s; P=0.489). The number of times the forceps were opened and closed during stone-catching was significantly lower with the one-surgeon technique than with the two-surgeon technique (5.0 [interquartile range 4.0-5.0] times vs 9.0 [interquartile range 6.0-14.5] times; P=0.018). The one-surgeon technique can offer comparable efficacy to the two-surgeon technique. If surgeons are sufficiently trained, a one-surgeon basketing technique for performing flexible ureteroscopy might become a new technical modality for extracting renal stones.

Research on the on-demand scheduling algorithm of intelligent routing load based on SDN

Aiming at the traditional on-demand scheduling method of routing load, there are many problems, such as long time to complete the total task and high energy consumption. An on-demand scheduling method of routing load based on multiple SINKs is proposed. The total routing load task is sent from multiple SINK nodes to SDN network. After receiving the sub-tasks sent by SINK, the intra-cluster nodes begin to collect network data together. After the cluster heads fuse the network data transmitted by the nodes, the results are sent to SINK nodes. As a result, the shortest total task completion time is obtained, and the shortest total task completion time is used as the objective function of routing load scheduling on demand. Ant colony algorithm is used to solve the problem and complete the scheduling. Experimental results show that the proposed method has shorter completion time and lower energy consumption of total task.