DiffPD: Differentiable Projective Dynamics

We present a novel, fast differentiable simulator for soft-body learning and control applications. Existing differentiable soft-body simulators can be classified into two categories based on their time integration methods: Simulators using explicit timestepping schemes require tiny timesteps to avoid numerical instabilities in gradient computation, and simulators using implicit time integration typically compute gradients by employing the adjoint method and solving the expensive linearized dynamics. Inspired by Projective Dynamics ( PD ), we present Differentiable Projective Dynamics ( DiffPD ), an efficient differentiable soft-body simulator based on PD with implicit time integration. The key idea in DiffPD is to speed up backpropagation by exploiting the prefactorized Cholesky decomposition in forward PD simulation. In terms of contact handling, DiffPD supports two types of contacts: a penalty-based model describing contact and friction forces and a complementarity-based model enforcing non-penetration conditions and static friction. We evaluate the performance of DiffPD and observe it is 4–19 times faster compared with the standard Newton’s method in various applications including system identification, inverse design problems, trajectory optimization, and closed-loop control. We also apply DiffPD in a reality-to-simulation ( real-to-sim ) example with contact and collisions and show its capability of reconstructing a digital twin of real-world scenes.

Genetic Algorithm-Based Trajectory Optimization for Digital Twin Robots

Mobile robots have an important role in material handling in manufacturing and can be used for a variety of automated tasks. The accuracy of the robot’s moving trajectory has become a key issue affecting its work efficiency. This paper presents a method for optimizing the trajectory of the mobile robot based on the digital twin of the robot. The digital twin of the mobile robot is created by Unity, and the trajectory of the mobile robot is trained in the virtual environment and applied to the physical space. The simulation training in the virtual environment provides schemes for the actual movement of the robot. Based on the actual movement data returned by the physical robot, the preset trajectory of the virtual robot is dynamically adjusted, which in turn enables the correction of the movement trajectory of the physical robot. The contribution of this work is the use of genetic algorithms for path planning of robots, which enables trajectory optimization of mobile robots by reducing the error in the movement trajectory of physical robots through the interaction of virtual and real data. It provides a method to map learning in the virtual domain to the physical robot.

Proactive Guidance for Accurate UAV Landing on a Dynamic Platform: A Visual–Inertial Approach

This work aimed to develop an autonomous system for unmanned aerial vehicles (UAVs) to land on moving platforms such as an automobile or a marine vessel, providing a promising solution for a long-endurance flight operation, a large mission coverage range, and a convenient recharging ground station. Unlike most state-of-the-art UAV landing frameworks that rely on UAV onboard computers and sensors, the proposed system fully depends on the computation unit situated on the ground vehicle/marine vessel to serve as a landing guidance system. Such a novel configuration can therefore lighten the burden of the UAV, and the computation power of the ground vehicle/marine vessel can be enhanced. In particular, we exploit a sensor fusion-based algorithm for the guidance system to perform UAV localization, whilst a control method based upon trajectory optimization is integrated. Indoor and outdoor experiments are conducted, and the results show that precise autonomous landing on a 43 cm × 43 cm platform can be performed.

Joint Multi-domain Resource Allocation and Trajectory Optimization in UAV-assisted M-IoT Networks

<div>The integration of Maritime Internet of Things (M-IoT) technology and unmanned aerial/surface vehicles (UAVs/USVs) has been emerging as a promising navigational information technique in intelligent ocean systems. With the unprecedented increase of computation-intensive yet latency sensitive marine mobile Internet services, mobile edge computing (MEC) and non-orthogonal multiple access (NOMA) have been envisioned as promising approaches to providing with the low-latency as well as reliable computing services and ultra-dense connectivity. In this paper, we investigate the energy consumption minimization based energy-efficient MEC via cooperative NOMA for the UAV-assisted M-IoT networks. We consider that USVs offload their computation-workload to the UAV equipped with the edge-computing server subject to the UAV mobility. To improve the energy efficiency of offloading transmission and workload computation, we focus on minimizing the total energy consumption by jointly optimizing the USVs’ offloaded workload, transmit power, computation resource allocation as well as the UAV trajectory subject to the USVs’ latency requirements. Despite the nature of mixed discrete and non-convex programming of the formulated problem, we exploit the vertical decomposition and propose a two-layered algorithm for solving it efficiently. Specifically, the top-layered algorithm is proposed to solve the problem of optimizing the UAV trajectory based on the idea of Deep Reinforcement Learning (DRL), and the underlying algorithm is proposed to optimize the underlying multi-domain resource allocation problem based on the idea of the Lagrangian multiplier method. Numerical results are provided to validate the effectiveness of our proposed algorithms as well as the performance advantage of NOMA-enabled computation offloading in terms of overall energy consumption.</div>

Joint Multi-domain Resource Allocation and Trajectory Optimization in UAV-assisted M-IoT Networks

<div>The integration of Maritime Internet of Things (M-IoT) technology and unmanned aerial/surface vehicles (UAVs/USVs) has been emerging as a promising navigational information technique in intelligent ocean systems. With the unprecedented increase of computation-intensive yet latency sensitive marine mobile Internet services, mobile edge computing (MEC) and non-orthogonal multiple access (NOMA) have been envisioned as promising approaches to providing with the low-latency as well as reliable computing services and ultra-dense connectivity. In this paper, we investigate the energy consumption minimization based energy-efficient MEC via cooperative NOMA for the UAV-assisted M-IoT networks. We consider that USVs offload their computation-workload to the UAV equipped with the edge-computing server subject to the UAV mobility. To improve the energy efficiency of offloading transmission and workload computation, we focus on minimizing the total energy consumption by jointly optimizing the USVs’ offloaded workload, transmit power, computation resource allocation as well as the UAV trajectory subject to the USVs’ latency requirements. Despite the nature of mixed discrete and non-convex programming of the formulated problem, we exploit the vertical decomposition and propose a two-layered algorithm for solving it efficiently. Specifically, the top-layered algorithm is proposed to solve the problem of optimizing the UAV trajectory based on the idea of Deep Reinforcement Learning (DRL), and the underlying algorithm is proposed to optimize the underlying multi-domain resource allocation problem based on the idea of the Lagrangian multiplier method. Numerical results are provided to validate the effectiveness of our proposed algorithms as well as the performance advantage of NOMA-enabled computation offloading in terms of overall energy consumption.</div>

Mediating Between Contact Feasibility and Robustness of Trajectory Optimization Through Chance Complementarity Constraints

As robots move from the laboratory into the real world, motion planning will need to account for model uncertainty and risk. For robot motions involving intermittent contact, planning for uncertainty in contact is especially important, as failure to successfully make and maintain contact can be catastrophic. Here, we model uncertainty in terrain geometry and friction characteristics, and combine a risk-sensitive objective with chance constraints to provide a trade-off between robustness to uncertainty and constraint satisfaction with an arbitrarily high feasibility guarantee. We evaluate our approach in two simple examples: a push-block system for benchmarking and a single-legged hopper. We demonstrate that chance constraints alone produce trajectories similar to those produced using strict complementarity constraints; however, when equipped with a robust objective, we show the chance constraints can mediate a trade-off between robustness to uncertainty and strict constraint satisfaction. Thus, our study may represent an important step towards reasoning about contact uncertainty in motion planning.