2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)最新文献_第5页

Embedded Barometric Pressure Sensor Unit for Force Myography in Exoskeletons 用于外骨骼力肌成像的嵌入式气压传感器单元

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000204

Charlotte Marquardt, Pascal Weiner, M. Dežman, T. Asfour

{"title":"Embedded Barometric Pressure Sensor Unit for Force Myography in Exoskeletons","authors":"Charlotte Marquardt, Pascal Weiner, M. Dežman, T. Asfour","doi":"10.1109/Humanoids53995.2022.10000204","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000204","url":null,"abstract":"Exoskeletons and other wearable devices benefit from sensor systems that are based on biofeedback for detecting muscle activity. However, such sensor systems in wearable applications are often impractical due to the need for direct skin contact or due to non-linear signal output, both requiring extensive calibration and post-processing. This paper presents a compact and robust barometer-based pressure sensor unit for detecting surface muscle pressure in an exoskeleton interface shell that is light, thin and does not require direct contact with the skin. It consists of an array of five miniature barometric sensors mounted on a custom embedded printed circuit board enclosed in a silicon dome. Evaluation of this sensor unit in a controlled experimental setup showed high sensitivity and an almost linear response to the normal force applied to the silicon dome. Additionally, a pilot study was conducted with four participants in a wearable application to compare the performance of the proposed sensor with that of electromyography (EMG). The results showed higher robustness to positioning, consistent signal, low variance and inter-subject variability compared to EMG.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"79 2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132726438","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Evaluation of Flexibility in Frontal Plane and Improvement of Usability for Dorsiflexion Support Unit 背屈支撑单元正面柔韧性评价与可用性提高

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000100

Takehito Kikuchi, Maki Nakahara, Toma Ono, Isao Abe

{"title":"Evaluation of Flexibility in Frontal Plane and Improvement of Usability for Dorsiflexion Support Unit","authors":"Takehito Kikuchi, Maki Nakahara, Toma Ono, Isao Abe","doi":"10.1109/Humanoids53995.2022.10000100","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000100","url":null,"abstract":"Aging society has recently been a serious social problem in many countries, and the frail elderly are focused on as major targets for human support devices. To support the dorsal flexion of their ankles, we have developed a dorsiflexion support unit (DSU) using elastomer-embedded flexible joints (EEFJ). In this study, we conducted gait analyses on cross-slope surfaces. By considering the results of the cross-walking experiments, the mechanical analyses are conducted with computer simulations. In addition, the usability tests for the DSU with an improved elastomer cap were conducted. The results of the gait analyses show the ankles maintain inversion without returning to their neutral positions in swing phases. This information is useful for designing orthosis, prostheses, and robotic gait aids. The results of the mechanical analyses show the required forces were 5 N for inversion, 7 N for eversion at 10 degrees rotations, and 10 N for inversion, 15 N for eversion at 20 degrees rotations. These asymmetric mechanical properties come from the non-linearity of the EEFJ. We improved the shape of the elastomer cap from a double flange type to a single flange type with a fixing bolt to reduce a burden of wearing. According to the results of usability tests, though the setting time was slightly increased, the rotational motions required for the settings were decreased by half for all participants.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"8 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124318915","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

TDE2-MBRL: Energy-exchange Dynamics Learning with Task Decomposition for Spring-loaded Bipedal Robot Locomotion 基于任务分解的弹簧双足机器人运动能量交换动力学学习

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000180

Cheng-Yu Kuo, Hirofumi Shin, Takumi Kamioka, Takamitsu Matsubara

{"title":"TDE2-MBRL: Energy-exchange Dynamics Learning with Task Decomposition for Spring-loaded Bipedal Robot Locomotion","authors":"Cheng-Yu Kuo, Hirofumi Shin, Takumi Kamioka, Takamitsu Matsubara","doi":"10.1109/Humanoids53995.2022.10000180","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000180","url":null,"abstract":"Spring-loaded Inverted Pendulum (SLIP) inspired bipedal robots (SLIP-biped) have high agility owing to their fault tolerance under impacts. Controlling a SLIP-biped requires capturing its dynamics; however, its high complexity makes analytic method implementation challenging. Thus, a Model-based Reinforcement Learning (MBRL) that learns a dynamics model and utilizes it for control design appears to be a reasonable alternative. Nevertheless, modeling high complexity dynamics with conventional MBRL approaches requires enormous samples or a high computation load. Therefore, exploring a simplified and compact dynamics model for SLIP-biped would be a key to increasing the feasibility of MBRL implementation and real-time control. We propose a Task-Decomposed Energy-exchange dynamics learning with MBRL (TDE2-MBRL) to capture simplified SLIP-biped dynamics and utilize them for control. Specifically, under the law of energy conservation, we model the energy exchange to reduce dynamics' dimensionality. Next, we decompose the SLIP-biped dynamics into locomotion task phases to cope with dynamics dissimilarity. The effectiveness is demonstrated by hopping skill acquisition with a precise simulated SLIP-biped replica of a real SLIP-biped. The experiment results show that TDE2-MBRL improves learning efficiency and control frequency while having comparable model accuracy to the standard MBRL.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115533517","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 2

Continuous Jumping of a Parallel Wire-Driven Monopedal Robot RAMIEL Using Reinforcement Learning 基于强化学习的平行线驱动单足机器人RAMIEL连续跳跃

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000182

Kento Kawaharazuka, Temma Suzuki, K. Okada, M. Inaba

{"title":"Continuous Jumping of a Parallel Wire-Driven Monopedal Robot RAMIEL Using Reinforcement Learning","authors":"Kento Kawaharazuka, Temma Suzuki, K. Okada, M. Inaba","doi":"10.1109/Humanoids53995.2022.10000182","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000182","url":null,"abstract":"We have developed a parallel wire-driven monope-dal robot, RAMIEL, which has both speed and power due to the parallel wire mechanism and a long acceleration distance. RAMIEL is capable of jumping high and continuously, and so has high performance in traveling. On the other hand, one of the drawbacks of a minimal parallel wire-driven robot without joint encoders is that the current joint velocities estimated from the wire lengths oscillate due to the elongation of the wires, making the values unreliable. Therefore, despite its high performance, the control of the robot is unstable, and in 10 out of 16 jumps, the robot could only jump up to two times continuously. In this study, we propose a method to realize a continuous jumping motion by reinforcement learning in simulation, and its application to the actual robot. Because the joint velocities oscillate with the elongation of the wires, they are not used directly, but instead are inferred from the time series of joint angles. At the same time, noise that imitates the vibration caused by the elongation of the wires is added for transfer to the actual robot. The results show that the system can be applied to the actual robot RAMIEL as well as to the stable continuous jumping motion in simulation.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131292652","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Development of a Multi-Fingered Hand with a Multi-Step Locking Mechanism for Carrying Heavy Objects by a Humanoid Robot 仿人机器人多指多步锁定机械手的研制

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000142

Yutaro Matsuura, Naoki Hiraoka, Kunio Kojima, Iori Yanokura, Hiroki Yoshioka, K. Okada, M. Inaba

{"title":"Development of a Multi-Fingered Hand with a Multi-Step Locking Mechanism for Carrying Heavy Objects by a Humanoid Robot","authors":"Yutaro Matsuura, Naoki Hiraoka, Kunio Kojima, Iori Yanokura, Hiroki Yoshioka, K. Okada, M. Inaba","doi":"10.1109/Humanoids53995.2022.10000142","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000142","url":null,"abstract":"Along with the progress of robot technology, robots are expected to work at construction and disaster sites. At such sites, multi-fingered hands are required to have the ability to manipulate objects of various shapes and a high endurance for supporting heavy objects. In this study, we propose a multi-step locking mechanism using cams for finger joints and a strategy for switching between locking and non-locking depending on target motions. This mechanism enabled a hand with the high back drivability for flexible manipulation and the endurance to continuously exert high torque regardless of the actuator's torque limit. By mounting the developed hand on a high-power and high-degree-of-freedom humanoid, the robot was able to perform high-load manipulation, such as pull-ups and bar-hanging, and tasks that require dexterity, such as tank-carrying operations. Through these experiments, we demonstrated that by combining the switching of the hand-locking mechanism with the dual-arm manipulation, the robot can perform tasks that require high manipulation force and dexterity. We found that a locking mechanism that can be turned on and off for each arm or each task is important for a dual-armed high-power humanoid robot.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"283 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124523586","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 1

Avatar-Darwin a Social Humanoid with Telepresence Abilities Aimed at Embodied Avatar Systems 阿凡达-达尔文是一种具有远程呈现能力的社会性类人，目标是具体化的阿凡达系统

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000176

Akshay Dave, Jean Chagas Vaz, Jeongeun Kim, Nicolas Kosanovic, Nathan Kassai, Paul Y. Oh

{"title":"Avatar-Darwin a Social Humanoid with Telepresence Abilities Aimed at Embodied Avatar Systems","authors":"Akshay Dave, Jean Chagas Vaz, Jeongeun Kim, Nicolas Kosanovic, Nathan Kassai, Paul Y. Oh","doi":"10.1109/Humanoids53995.2022.10000176","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000176","url":null,"abstract":"Human compassion has driven the creation of social robots that provide aid and interactions to people in need. However, such systems are preprogrammed and are therefore asocial as they lack adaptability and authenticity in their interactions. In this paper, the miniature biped DARwIn-OP was upgraded into Avatar-DARwIn, a humanoid telepresence platform that broadcasts human cognition. To address a body correspondence problem related to miniature humanoid embodiment, a “Marionette Method” of scaling was developed. Furthermore, results have shown that operator motions were successfully retargeted to Avatar-DARwIn, allowing operators to complete an array of social tasks. Additionally, a System Usability Scale (SUS) Test was administered amongst a pool of eleven participants to evaluate operator experience and system functionality. Data collected from SUS testing (based on a 5-point LIKERT scale) found that operators felt a high sense of telepresence as this category received a score of 4.36. Overall, this teleoperated, real-time telepresence system allows human operators to embody a miniature social humanoid to connect people around the world, regardless of distance.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"600 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123146557","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

An MPC-Based Framework for Dynamic Trajectory Re-Planning in Uncertain Environments 不确定环境下基于mpc的动态轨迹重规划框架

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000159

Maolin Lei, Liang Lu, Arturo Laurenzi, Luca Rossini, Edoardo Romiti, J. Malzahn, N. Tsagarakis

{"title":"An MPC-Based Framework for Dynamic Trajectory Re-Planning in Uncertain Environments","authors":"Maolin Lei, Liang Lu, Arturo Laurenzi, Luca Rossini, Edoardo Romiti, J. Malzahn, N. Tsagarakis","doi":"10.1109/Humanoids53995.2022.10000159","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000159","url":null,"abstract":"Online motion re-planning is an important feature for introducing robots into unstructured environments where the close presence of humans at any time can challenge the operation of the robot from the human safety perspective. This work introduces a novel re-planning framework for robotic manipulators operating in dynamic environments where the interactions with humans may occur either in an anticipated or unexpected manner. The contribution of the proposed framework lies in the fact that it allows to account for the uncertainty of human pose and challenges associated with human motion estimation during occlusion phases of the human with respect to the perception system on the robot. To this aim the proposed framework is comprised of an uncertainty estimation component and a model predictive control (MPC) component, the combination of which enables to efficiently and dynamically track a task-space trajectory by the robot while limiting the probability of potential collisions with a moving human obstacle entering the workspace of the robot. Simulations and experimental trials on a robotic platform show the effectiveness of the proposed framework in re-planning the trajectory of the robotic arm under the presence of a human detected by the perception system installed on the robot.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"213 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123297299","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 2

Minimalistic Soft Exosuit for Assisting the Shoulder via Biomechanics-Aware Optimization 通过生物力学感知优化来辅助肩部的极简软外套

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000128

Sagar Joshi, Irene Beck, A. Seth, C. D. Santina

{"title":"Minimalistic Soft Exosuit for Assisting the Shoulder via Biomechanics-Aware Optimization","authors":"Sagar Joshi, Irene Beck, A. Seth, C. D. Santina","doi":"10.1109/Humanoids53995.2022.10000128","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000128","url":null,"abstract":"Soft exosuits can help to prevent work-related musculoskeletal disorders by offloading human muscles through the application of external forces across the human joints. Many exosuits achieve this through tension producing elements called as exotendons. However, the design of these devices is based on intuition and experience. This leads to potentially sub-optimal or even harmful designs that could cause discomfort or injury to the wearer. This paper deals with automatically finding appropriate attachments and routing locations for exotendons. We propose to do that by accurate musculoskeletal modeling and design parameter optimization of soft exosuits. We focus here on a soft exosuit with a single passive exotendon to assist the shoulder. Using three arm raising-lowering and internal-external rotation motions as examples, we optimize the attachment point and rest-length of the exotendon to reduce overall muscle effort. We then fabricate the exosuit and validate the model predictions by testing with six participants. Supporting the predictions from simulations, measured muscle activity shows reductions in the deltoid and trapezius muscles. This work represents the first instance of explicitly optimizing functional and geometric parameters of exotendons in wearable assistive devices for minimizing human effort.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130114699","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Object-Centric Grasping Transferability: Linking Meshes to Postures 以对象为中心的抓取可转移性:将网格连接到姿势

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000192

Diego Hidalgo-Carvajal, Carlos Magno Catharino Olsson Valle, Abdeldjallil Naceri, S. Haddadin

{"title":"Object-Centric Grasping Transferability: Linking Meshes to Postures","authors":"Diego Hidalgo-Carvajal, Carlos Magno Catharino Olsson Valle, Abdeldjallil Naceri, S. Haddadin","doi":"10.1109/Humanoids53995.2022.10000192","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000192","url":null,"abstract":"Attaining human hand manipulation capabilities is a sought-after goal of robotic manipulation. Several works have focused on understanding and applying human manipulation insights in robotic applications. However, few considered objects as central pieces to increase the generalization properties of existing methods. In this study, we explore context-based grasping information transferability between objects by using mesh-based object representations. To do so, we empirically labeled, in a mesh point-wise manner, 10 grasping postures onto a set of 12 purposely selected objects. Subsequently, we trained our convolutional neural network (CNN) based architecture with the mesh representation of a single object, associating grasping postures to its local regions. We tested our network across multiple objects of distinct similarity values. Results show that our network can successfully estimate non-feasible grasping regions as well as feasible grasping postures. Our results suggest the existence of an abstract relation between the predicted context-based grasping postures and the geometrical properties of both the training and test objects. Our proposed approach aims to expand grasp learning research by linking local segmented meshes to postures. Such a concept can be applied to grasp new objects using anthropomorphic robot hands.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"9 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128335417","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 2

Efficient Transformers for on-robot Natural Language Understanding 面向机器人自然语言理解的高效转换器

2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids) Pub Date : 2022-11-28 DOI: 10.1109/Humanoids53995.2022.10000085

Antonio Greco, Antonio Roberto, Alessia Saggese, M. Vento

{"title":"Efficient Transformers for on-robot Natural Language Understanding","authors":"Antonio Greco, Antonio Roberto, Alessia Saggese, M. Vento","doi":"10.1109/Humanoids53995.2022.10000085","DOIUrl":"https://doi.org/10.1109/Humanoids53995.2022.10000085","url":null,"abstract":"The main task of a social robot is to interact with humans through spoken natural language. It implies that it must be able to understand the intent of the user and the involved entities. Recently, different solutions have been proposed to deal with the Natural Language Understanding (NLU) task. Extremely accurate results have been obtained by architectures based on transformers, but they require high computational resources to work in real-time. Unfortunately, these resources are not available on embedded systems equipped on board the robot. For these reasons, in this paper we experimentally evaluate the most promising transformers for NLU over the popular ATIS and SNIPS datasets and measured their inference time on the NVIDIA Jetson Xavier NX embedded system. The experimental analysis demonstrates that the Albert model can obtain comparable performance w.r.t. the popular BERT architecture (just a 2% drop on entity recognition), while gaining a speed-up of more than 3x. Thanks to the insights coming out from our analysis, we finally developed a real system for restaurant search running the model over a NVIDIA Jetson Xavier NX equipped on board of a social robot, obtaining a positive user feedback about its effectiveness and responsiveness.","PeriodicalId":180816,"journal":{"name":"2022 IEEE-RAS 21st International Conference on Humanoid Robots (Humanoids)","volume":"215 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116359426","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0