The integration of AI tasks into Artificial Intelligence of Things (AIoT) devices has gained significant attention for its ability to reduce transmission latency and enhance data privacy compared to traditional cloud-based solutions. However, heterogeneous AIoT platforms face the challenge of balancing stringent latency constraints with energy efficiency under dynamic AI workloads. To address this issue, we propose BOLT-PM, a Bayesian Optimization framework for Latency-sensitive Task and Power Management, which dynamically allocates computing resources and optimizes power consumption in AIoT devices. We design a multi-branch neural network to accurately predict task execution times under varying resource constraints, enabling precise and adaptive performance estimation. We further enhance the Bayesian optimization process by embedding a variational autoencoder (VAE) to construct a smooth latent representation of the search space, thereby accelerating convergence and improving optimization stability. We implement BOLT-PM as a lightweight, platform-agnostic runtime framework that continuously adapts to workload fluctuations, maintaining latency guarantees while minimizing energy consumption. We evaluate BOLT-PM on several commercial AIoT boards, including RK3588, Jetson TX2, and Raspberry Pi, and compare it against classical heuristic methods and state-of-the-art approaches. Experimental results show that BOLT-PM achieves substantial energy savings while ensuring low-latency AI task execution, demonstrating its effectiveness as a robust and energy-efficient solution for power-aware AIoT applications.

Mobile manipulators enable a wide range of operations with mobility and advanced manipulation capabilities. Despite their potential, existing approaches typically treat the mobile base and the manipulator separately, thereby limiting the optimality of the system for composite whole-body behaviors. In this work, we present a Whole-Body Model Predictive Control framework for mobile manipulation involving tasks with varying timelines. We integrate task priorities across both task and time dimensions, bringing inherent transition ability with enhanced performance. Our approach improves the trajectory tracking performance by up to 36 % in terms of manipulability and reduces the maximum velocity during task priority transitions by 53% compared to the existing approach while maintaining a low computational cost of 4.3 ms, allowing for high reactivity in real-world applications. We demonstrate its effectiveness through a door-opening and traversing behavior, showcasing the first successful implementation of a non-holonomic mobile manipulator in such a scenario. See https://wbmpc.github.io/ for supplemental materials.

Humanoid robots encounter considerable difficulties in autonomously recovering from falls, especially within dynamic and unstructured environments. Conventional control methodologies are often inadequate in addressing the complexities associated with high-dimensional dynamics and the contact-rich nature of fall recovery. Meanwhile, reinforcement learning techniques are hindered by issues related to sparse rewards, intricate collision scenarios, and discrepancies between simulation and real-world applications. In this study, we introduce a multi-stage curriculum learning framework, termed HiFAR. This framework employs a staged learning approach that progressively incorporates increasingly complex and high-dimensional recovery tasks, thereby facilitating the robot’s acquisition of efficient and stable fall recovery strategies. Furthermore, it enables the robot to adapt its policy to effectively manage real-world fall incidents. We assess the efficacy of the proposed method using a real humanoid robot, showcasing its capability to autonomously recover from a diverse range of falls with high success rates, rapid recovery times, robustness, and generalization.

With the rapid growth of IoT technology and the rise of smart devices, edge computing, particularly federated learning (FL), has gained importance for preserving user data privacy. However, FL faces challenges like non-independent identically distributed data and device heterogeneity, leading to model disparities and reduced precision. Our research proposes a novel adaptive FL framework specifically engineered to synchronize computational capacities within heterogeneous edge computing landscapes. Building upon the proof of convergence boundaries for local aggregation model, this algorithm adapts the number of iterations for local updates by considering the resource consumption relationship between local aggregation model and the local updated model by various clients. This method exhibit adaptability within an environment where disparities in edge device computational capacities exist, effectively balancing computational prowess among diverse devices and enhancing the output performance of federated learning Experiments on MNIST and PlantVillage datasets show that in heterogeneous environments, our algorithm outperforms existing methods, improving the loss function by at least 16.87% and the convergence speed by at least 2 times, in various environments (MobileNet, AlexNet).

The robustness of legged locomotion is crucial for quadrupedal robots in challenging terrains. Recently, Reinforcement Learning (RL) has shown promising results in legged locomotion and various methods try to integrate privileged distillation, scene modeling, and external sensors to improve the generalization and robustness of locomotion policies. However, these methods are hard to handle uncertain scenarios such as abrupt terrain changes or unexpected external forces. In this paper, we consider a novel risk-sensitive perspective to enhance the robustness of legged locomotion. Specifically, we employ a distributional value function learned by quantile regression to model the aleatoric uncertainty of environments, and perform risk-averse policy learning by optimizing the worst-case scenarios via a risk distortion measure. Extensive experiments in both simulation environments and a real Aliengo robot demonstrate that our method is efficient in handling various external disturbances, and the resulting policy exhibits improved robustness in harsh and uncertain situations in legged locomotion.

Despite the impressive performance of Reinforcement Learning (RL), the black-box neural network backbone hinders users from trusting and deploying trained agents in real-world applications where safety is crucial. In order to make agents more trustworthy and controllable, for a given RL-trained policy, we propose to enhance its interpretability and make it human-controllable without retraining. We accomplish this goal by following a 3-step pipeline: 1) We interpret neurons by analyzing the causal effect of neurons on the kinematic attributes; To help agents unlock novel skills and enable human to assist agents in accomplishing tasks, 2) we locate the X-neuron, the optimal neuron that is capable of evoking the desired behavior; 3) and edit its activation values to achieve the precise control. We evaluate our method on various RL tasks ranging from autonomous driving to robot locomotion, and the results display that our approach outperforms previous work regarding almost all evaluation metrics. Through enhancing interpretability and introducing human control, the agents can improve safety and performance, even in unseen environments and novel tasks. For locomotion robots simply trained to walk forward, our method unlocks diverse controllable behaviors ranging from jump to backflip.

For the roll control of an unmanned motorcycle equipped with twin control moment gyros (CMGs), the optimal allocation between the steering motor and CMGs when encountering strong disturbance is rarely investigated in existing studies. In this paper, based on an extended state observer (ESO) and a novel control allocator (CA), a robust control scheme ESO-CA is proposed for the unmanned motorcycle to resist disturbances. The ESO enables the estimation and compensation of the lumped disturbance, which incorporates unmodeled dynamics, parameter perturbations and external disturbances. Based on a novel optimization formulation that considers the integral saturations of actuators, the control allocation method is developed to minimize total energy consumption and reduce steering chattering. Two comparative numerical simulations and one physical experiment are provided to demonstrate the effectiveness of the proposed control scheme.

In this article, we study the problem of designing an adaptive scheduling scheme for dynamic multiple-criticality real-time applications in vehicular edge computing systems. This scheduling problem is formulated as a mixed-integer nonlinear problem. We propose a workload-aware scheduling approach that not only guarantees the applicaitons' mixed-criticality schedulability but also adaptively manages their execution depending on their released frequencies at runtime. In particular, we first present the response time analysis for multiple-criticality applications in the edge computing system with different computing capability servers. Then, we derive a state-transition equation that makes the dynamic programming applicable to building an excellent schedule for one specific criticality-level mode. Such an approach is extended to the system's different criticality-level modes. For the purpose of increasing the quality-of-service toward low-critical applications, we leave low-critical applications to execute as much as possible at runtime, depending on their predicted frequencies. Extensive experimental results show the superiority of our proposed approaches in terms of improving the schedulability success rate and reducing the number of suspended applications online.

Due to the flexibility, availability, and scalability of cloud computing services, more and more users seek solutions via cloud computing techniques. A cloud computing platform often consists of a large number of infrastructures, and its energy consumption is a big problem. In this article, we study how to minimize the energy consumption of a cloud computing platform when handling some intermittent real-time tasks. Unlike previous works that abstract users’ submitted tasks as single computation jobs and process them using CPU, this work proposes using CPU and GPU to process intermittent real-time tasks that occur at irregular intervals and their released computation jobs must be completed within required time limits. The energy consumption minimization problem is formulated as an integer nonlinear programming problem that needs to decide on a task assignment plan and a specific resource allocation plan. To effectively solve this problem, we define a state that represents the optimal solution for a given set of tasks with a given amount of resources, as well as a value function that represents the value of a state. In this way, we derive a state-transition equation and develop a dynamic programming method to solve the problem. This method is also extended to handle tasks whose arrival time is dynamic and unpredictable. Experiments show that the proposed algorithm can effectively reduce energy consumption, while its computation time is quite low compared to some other greedy methods.

The hierarchical quadratic programming (HQP) is commonly applied to consider strict hierarchies of multi-tasks and robot's physical inequality constraints during whole-body compliance. However, for the one-step HQP, the solution can oscillate when it is close to the boundary of constraints. It is because the abrupt hit of the bounds gives rise to unrealizable jerks and even infeasible solutions. This paper proposes the mixed control, which blends the single-axis model predictive control (MPC) and proportional derivative (PD) control for the whole-body compliance to overcome these deficiencies. The MPC predicts the distances between the bounds and the control target of the critical tasks, and it provides smooth and feasible solutions by prediction and optimization in advance. However, applying MPC will inevitably increase the computation time. Therefore, to achieve a 500 Hz servo rate, the PD controllers still regulate other tasks to save computation resources. Also, we use a more efficient null space projection (NSP) whole-body controller instead of the HQP and distribute the single-axis MPCs into four CPU cores for parallel computation. Finally, we validate the desired capabilities of the proposed strategy via simulations and the experiment on the humanoid robot Walker X.

Whole-body control (WBC) with task priority transition is an important technology for robots to switch multiple behaviors, change different objectives, and adapt to various environments. Many methods have solved the problem of control continuity in the priority transition process. However, they either increased the computation consumption or sacrificed the accuracy of tasks in practical application. In this work, we propose a Recursive Hierarchical Projection (RHP) matrix and introduce it in Hierarchical Quadratic Programming (HQP). This RHP-HQP scheme can form continuously changing hierarchical projection and regard the WBC problem with task priority transition as a unified formulation. This unified formulation can be smoothly transitioned without increasing computation consumption and solved without losing task accuracy. The comparative simulations of the reactive collision avoidance verify that this priority transition scheme can guarantee high computational efficiency and task accuracy.

For achieving kinematically suitable configurations and highly dynamic task execution, an efficient way is to consider robot performance indices in the whole-body control (WBC) of robots. However, current WBC methods have not considered the intrinsic features of parallel robots, especially motion/force transmissibility (MFT). This paper proposes an MFT-enhanced WBC scheme for parallel-legged robots. Introducing the performance indices of MFT into a WBC is challenging due to the nonlinear relationship between MFT indices and the robot configuration. To overcome this challenge, we establish the MFT preferable space of the robot offline and formulate it as a polyhedron in the joint space at the acceleration level. Then, the WBC employs the polyhedron as a soft constraint. As a result, the robot possesses high-speed and high-acceleration capabilities by satisfying this constraint. The offline preprocessing relieves the online computation burden and helps the WBC achieve a 1kHz servo rate. Finally, we validate the performance and robustness of the proposed method via simulations and experiments on a parallel-legged bipedal robot.

This work considers the control of underactuated bipedal walking, and a novel capturability-based control framework is presented. Compared with traditional approaches, the presented control method does not rely on the use of the Poincaré map, which may take significant computational cost. Firstly, a new definition of stable walking is presented, and a novel foot-placement based control method is proposed. Then, a controller design method is presented based on this control method. For the controller design, the foot placement adjustment is achieved by updating the virtual constraints using a heuristic method, and an improved virtual constraint control method is proposed to enforce the virtual constraints. Finally, the effectiveness of the presented control framework is illustrated on a five-link underactuated planar biped by numerical simulations.

Most researches on balance control of the bicycle robots are for the situation that the bicycle robot is with constant forward velocity and constant feedback gain, but are not appropriate to be employed for time varying forward velocity situation. In this paper, the nonlinear Euler-Lagrange model of the bicycle robot and the simplified nonlinear descriptor state space model are firstly deduced. A polynomial controller, rather than a constant gain feedback one, is proposed, which constituting the nonlinear closed-loop descriptor system. The sufficient condition on examining the stability of closed-loop system, together with one alternative method on designing a polynomial controller utilizing the SOSTool, are then proposed. By this work, the polynomial controller is broadened to be applied on nonlinear descriptor system, and on the balance and forward control of bicycle robot with time varying forward velocity. One numerical example shows the capacity of the control scheme proposed in this paper.

On a passive dynamic walker, we find that an appropriate spring placed on the mass center of two legs can greatly improve the walker's walking performance, which includes its walking speed, disturbance rejection ability and step length. However, the extent to which spring influences these properties is unknown and how to choose an appropriate spring stiffness to achieve optimal walking performance still needs to be discussed. In this paper, we present a study on the effect of spring on these properties by constructing a synthesize index P to assess the walking performance. Through numerical simulation on three models, we find the walker with spring has a better walking performance over a pure passive dynamic walker and the extension spring on the mass center of two legs can improve a walker's overall walking performance much better.

This paper presents our study over the effect of complementary energy feedback on virtual slope walking, while virtual slope walking is our new biped gait generation method inspired by passive dynamic walking. The energy feedback strength is defined and the walking is modeled as a step-to-step function. The Jacobi matrix eigenvalues of the function are calculated together with the basin of attraction. From the analysis, we find the characteristic of Complementary Energy Feedback is being effective on a fast gait but weak on a slow one. By making use of the complementary energy feedback in walking experiment, our robot achieves speed change from 1.5 leg/s to 4.1 leg/s.

In our previous work, we have realized virtual slope walking that a robot can walk on level ground as it walks down a virtual slope by leg length modulation. In this paper, we present the instantaneous leg extension model of virtual slope walking to analyze the essentials of virtual slope walking. It has two straight massless legs and a point mass body at the hip. The stance leg is extended instantaneously while the swing leg is swung and shortened actively. We demonstrate that this model can exhibit stable walking cycles on level ground. We obtain the sufficient conditions for the existence of the fixed point. We then illustrate the effect of the model parameters on the fixed point to show how the fixed point can be determined by adjusting the parameters. Further, we theoretically proved that the fixed point is asymptotically stable, meaning that it is independent on the initial conditions. The validity of the proposed model has been examined by numerical simulations.

This article presents an adaptive dynamic clustered particle filtering method for mobile robot global localization. The posterior distribution of robot pose in global localization is usually multimodal due to the symmetry of the environment and ambiguous detected features. Moreover, the multimodal distribution of the posterior varies as the robot moves and observations are obtained. Considering these characteristics, we use a set of clusters of particles to represent the posterior. These clusters are dynamically evolved corresponding to the varying posterior by merging the overlapping clusters and splitting the diffuse clusters or those whose particles gather to some sub-clusters inside. Further, in order to improve computational efficiency without sacrificing estimation accuracy, a mechanism for adapting the sample size of clusters is proposed. The theoretical lower bound of the number of particles needed to limit the estimation error is derived, based on the central limit theorem in multidimensional space and the statistic theory of ImportanceSampling (IS). Simulation results show the effectiveness of the proposed method, which is sufficient to achieve robust tracking of robot’s real pose and meanwhile significantly enhance the computational efficiency.

This paper presented a brief description for the gait planning of quadruped robot named Aibo ERS-7 which is a standard platform in the RoboCup 4-legged league. We approach a spline shaped locus to reduce the dimension of the parameter optimizing space and solve the problem of the significant bias between the planned locus and the real one. The result shows that the spline shaped locus is effective in finding the optimized locus shape in a short time. Finally, the robot achieves a gait faster than any previously known learned gait for Aibo

This paper proposes a design method of decentralized robust controllers for robots with joint backlash, flexibility and damping characteristics. For each joint subsystem, a robust tracking controller is designed in two steps: first, a nominal controller is designed for the nominal plant to get desired tracking performance, then a robust compensator is added to restrain the influence of the perturbation, that is the difference of the real plant from the nominal plant. The controller designed by the proposed method is a linear time-invariant one. It is shown that by applying the controller with a sufficiently wide frequency bandwidth robust output tracking property can be achieved in the contact phase while a new contact of the motor with the load in the correct direction is ensured in the backlash phase. An important feature of the method is that the controller parameters can be tuned on-line easily

Describes the progress of the control system design and implementation of the THBIP-I humanoid robot. The robot has 32 degrees of freedom and each joint is driven by a brushless DC electronic motor. Screw/nuts transmission mechanism is adapted in some joints of lower limbs to achieve compact and good dynamic performance. The control system of the robot has four subsystems: remote brain work station, mobile controller, distributed control units and sensor processing unit. At the present state, the lower limbs and upper limbs have been built and tested with off line gait planning. The distributed control units use PID schemes to servo the pre-generated joint trajectories. Under this architecture, the robot can perform stable walking with 30 centimeters step at 20 second per step.