Wearable exoskeleton robots have become a promising technology for supporting human motions in multiple tasks. Activity recognition in real-time provides useful information to enhance the robot’s control assistance for daily tasks. This work implements a real-time activity recognition system based on the activity signals of an inertial measurement unit (IMU) and a pair of rotary encoders integrated into the exoskeleton robot. Five deep learning models have been trained and evaluated for activity recognition. As a result, a subset of optimized deep learning models was transferred to an edge device for real-time evaluation in a continuous action environment using eight common human tasks: stand, bend, crouch, walk, sit-down, sit-up, and ascend and descend stairs. These eight robot wearer’s activities are recognized with an average accuracy of 97.35% in real-time tests, with an inference time under 10 ms and an overall latency of 0.506 s per recognition using the selected edge device.

Human Activity Recognition (HAR) has gained significant attention due to its broad range of applications, such as healthcare, industrial work safety, activity assistance, and driver monitoring. Most prior HAR systems are based on recorded sensor data (i.e., past information) recognizing human activities. In fact, HAR works based on future sensor data to predict human activities are rare. Human Activity Prediction (HAP) can benefit in multiple applications, such as fall detection or exercise routines, to prevent injuries. This work presents a novel HAP system based on forecasted activity data of Inertial Measurement Units (IMU). Our HAP system consists of a deep learning forecaster of IMU activity signals and a deep learning classifier to recognize future activities. Our deep learning forecaster model is based on a Sequence-to-Sequence structure with attention and positional encoding layers. Then, a pre-trained deep learning Bi-LSTM classifier is used to classify future activities based on the forecasted IMU data. We have tested our HAP system for five daily activities with two tri-axial IMU sensors. The forecasted signals show an average correlation of 91.6% to the actual measured signals of the five activities. The proposed HAP system achieves an average accuracy of 97.96% in predicting future activities.

Dual-arm robots can perform bimanual long-horizon (LH) manipulation, surpassing the capabilities of single-arm robots. However, bimanual LH tasks are challenging for robot intelligence due to the complexity of long sequence variables and multi-agent interactions. While Multi-Agent Reinforcement Learning (MARL) has shown promising results in agent interactions, these models struggle with sequential LH tasks due to limitations in credit assignment, vanishing memory, and the exploration-exploitation trade-off. This paper introduces a novel dual-arm robot intelligence framework, Temporal-Context Transformer Reinforcement Learning (TC-TRL), which integrates both a hybrid offline-online policy and imitation learning. TC-TRL leverages the attention mechanism to identify relevant temporal-context information from the LH observations space, updating the encoder value function and generating an optimal actions sequence using a decoder module, which uses demonstration guidance during online training. TC-TRL is tested on six bimanual tasks, and its performance is compared against five baseline RLs: MAPPO, HAPPO, IPPO, MAT, and DA-MAT. The results show that TC-TRL outperforms the three PPO-based RLs with an average success rate of 63.46%, 42.23% against MAT, and 30.91% for DA-MAT.

Dexterous object manipulation using anthropomorphic robot hands is of great interest for natural object manipulations across the areas of healthcare, smart homes, and smart factories. Deep reinforcement learning (DRL) is a particularly promising approach to solving dexterous manipulation tasks with five-fingered robot hands. Yet, controlling an anthropomorphic robot hand via DRL in order to obtain natural, human-like object manipulation with high dexterity remains a challenging task in the current robotic field. Previous studies have utilized some predefined human hand poses to control the robot hand’s movements for successful object-grasping. However, the hand poses derived from these grasping taxonomies are limited to a partial range of adaptability that could be performed by the robot hand. In this work, we propose a combinatory approach of a deep transformer network which produces a wider range of natural hand poses to configure the robot hand’s movements, and an adaptive DRL to control the movements of an anthropomorphic robot hand according to these natural hand poses. The transformer network learns and infers the natural robot hand poses according to the object affordance. Then, DRL trains a policy using the transformer output to grasp and relocate the object to the designated target location. Our proposed transformer-based DRL (T-DRL) has been tested using various objects, such as an apple, a banana, a light bulb, a camera, a hammer, and a bottle. Additionally, its performance is compared with a baseline DRL model via natural policy gradient (NPG). The results demonstrate that our T-DRL achieved an average manipulation success rate of 90.1% for object manipulation and outperformed NPG by 24.8%.

In this work, we present an approach of long-horizon intelligence that utilizes Sub-goal network based hierarchical reinforcement learning (HRL) for long-horizon tasks by a single-arm robot. Long-horizon (LH) tasks are complicated due to their longer complex sequences and the large number of environmental variables. We attempt to solve the LH learning problem by the Sub-goal network based HRL. The proposed approach is tested in both simulation and hardware environments by a LH task of opening a drawer, grasping and relocating an object, and closing a drawer. Our Sub-goal network based HRL achieves a success rate of 90.3% in completing the LH tasks. Whereas the conventional deep reinforcement learning solution could not complete the LH task.