With Moore’s law coming to its limits, the rate of increase in compute power available for processing applications is similarly coming to a halt. This implies that the compute intensive tasks, such as robotics, artif...
With Moore’s law coming to its limits, the rate of increase in compute power available for processing applications is similarly coming to a halt. This implies that the compute intensive tasks, such as robotics, artificial intelligence, and high-performance space computing need innovative ways to cater their ever-increasing compute needs. One innovative way to solve computational bottlenecks is to bring compute and memory together, as opposed to the Von Neumann computational model, with greater degree of parallelism in an event-based, asynchronous computation paradigm. Neuromorphic computing is one such paradigm that draws its inspiration from the brain. Energy and computational efficiency, asynchronous and event-based processing being its salient features, neuromorphic computing is an area worth exploring for compute intensive tasks. In this paper, the authors explore the possibilities and benefits of neuromorphic computing in robotics, and establish possible research directions that could benefit the robotics domain.
Modern manufacturing systems strive to optimize many different key performance indicators. Additionally, mathematical programming-based scheduling models enable the explicit considerations of constraints. However, due...
详细信息
Modern manufacturing systems strive to optimize many different key performance indicators. Additionally, mathematical programming-based scheduling models enable the explicit considerations of constraints. However, due to the ever-increasing demand for customized products it is not guaranteed that all constraints can be met at the same time. An approach to circumvent this problem of infeasibility is to replace the constraints by soft-constraints, i.e., to penalize their violation in the objective function. The presence of multiple competing objectives and soft-constraints gives rise to multi-objective optimization problems, where a decision maker has to weigh and balance the different objectives and soft-constraints according to his/her preferences and priorities. Ideally the values of all objectives and soft-constraints should lie within the same order of magnitude, making it easy to weigh them against each other. However, for heterogeneous objectives and soft-constraints with different scales and units this might be challenging. This paper presents an evolutionary algorithm for the optimal parametrization of multi-objective mixed-integer linear programming-based scheduling models. The goal of the evolutionary algorithm is to compute weights for the different terms of the objective which lead to a balanced influence on the overall objective at the optimum. Furthermore, the algorithm is extended by introducing an a priori weighing of the objectives in the fitness function of the evolutionary algorithm. The method is demonstrated on a scheduling problem in which a given set of orders has to be allocated to machines within a manufacturing environment.
Optical sorters separate particles of different classes by first detecting them while they are transported, e.g., on a conveyor belt, and subsequently bursting out particles of undesired classes using compressed air n...
详细信息
Optical sorters separate particles of different classes by first detecting them while they are transported, e.g., on a conveyor belt, and subsequently bursting out particles of undesired classes using compressed air nozzles. Currently, the most promising results are achieved by predictive tracking, a multitarget tracking approach based on extracted midpoints from area-scan camera images that analyzes the particles’ motion and activates the nozzles accordingly. However, predictive tracking requires expert knowledge for setup and preceding object detection. Moreover, particle shapes are only considered implicitly, and the need to solve an association problem rises the computational complexity of the algorithm. In this paper, we present GridSort, an image-based approach that forecasts the scene at the nozzle array using a convolutional long short-term memory neural network and subsequently extracts nozzle activations, thus circumventing the aforementioned weaknesses. We show how GridSort can be trained in an unsupervised fashion and evaluate it using a coupled discrete element–computational fluid dynamics simulation of an optical sorter. We compare our method with predictive tracking in terms of sorting accuracy and demonstrate that it is an easy-to-apply alternative while achieving state-of-the-art results.
Even with favorable policy frameworks, green hydrogen has not been cost-competitive in Europe. Optimizing hydrogen production can reduce the Levelized Costs of Hydrogen (LCOH). In this research, an electrolysis system...
详细信息
Wildland fires pose a terrifying natural hazard, underscoring the urgent need to develop datadriven and physics-informed digital twins for wildfire prevention, monitoring, intervention, and response. In this direction...
详细信息
The use of distributed optimization in machine learning can be motivated either by the resulting preservation of privacy or the increase in computational efficiency. On the one hand, training data might be stored acro...
详细信息
The use of distributed optimization in machine learning can be motivated either by the resulting preservation of privacy or the increase in computational efficiency. On the one hand, training data might be stored across multiple devices. Training a global model within a network where each node only has access to its confidential data requires the use of distributed algorithms. Even if the data is not confidential, sharing it might be prohibitive due to bandwidth limitations. On the other hand, the ever-increasing amount of available data leads to large-scale machine learning problems. By splitting the training process across multiple nodes its efficiency can be significantly increased. This paper aims to demonstrate how dual decomposition can be applied for distributed training of K-means clustering problems. After an overview of distributed and federated machine learning, the mixed-integer quadratically constrained programming-based formulation of the K-means clustering training problem is presented. The training can be performed in a distributed manner by splitting the data across different nodes and linking these nodes through consensus constraints. Finally, the performance of the subgradient method, the bundle trust method, and the quasi-Newton dual ascent algorithm are evaluated on a set of benchmark problems. The main benefit stemming from the formulation of the clustering problem as a mixed-integer program and from the use of dual decomposition within a federated learning framework, apart from the preservation of privacy, is the computation of a lower bound of the objective of the overall clustering problem. In this way, the worst-case distance of any found solution to the global optimum can be easily assessed. While the mixed-integer programming-based formulation of the clustering problems suffers from weak integer relaxations, the presented approach can potentially be used to enable an efficient solution in the future, both in a central and distributed sett
This paper presents a dual decomposition-based distributed optimization algorithm and applies it to distributed model predictive control (DMPC) problems. The considered DMPC problems are coupled through shared limited...
详细信息
ISBN:
(数字)9781665406734
ISBN:
(纸本)9781665406741
This paper presents a dual decomposition-based distributed optimization algorithm and applies it to distributed model predictive control (DMPC) problems. The considered DMPC problems are coupled through shared limited resources. Lagrangian duality can be used to decompose an MPC problem, so that each subsystem can compute its individual resource utilization, without sharing information, such as dynamics or constraints, with the other subsystems. The feasibility of the central problem is ensured by the coordination of the subproblems through dual variables which can be interpreted as prices on the shared limited resources. The proposed coordination algorithm makes efficient use of information collected from previous iterations by performing a quadratic approximation of the dual function of the central MPC problem. Aggressive update steps of the dual variables are prevented through a covariance-based step size constraint. The nonsmoothness encountered in dual optimization problems is addressed by the construction of cutting planes, similar to bundle methods for nonsmooth optimization. The cutting planes ensure that the updated dual variables do not lie outside the range of validity of the dual approximation. The proposed algorithm is evaluated on a two-tank system and compared to the standard subgradient method. The results show that the rate of convergence towards the centralized solution can be significantly improved while still preserving privacy between the subsystems through limited information exchange.
This paper presents a hierarchical distributed optimization algorithm based on quasi-Newton update steps. Separable convex optimization problems are decoupled through dual decomposition and solved in a distributed fas...
详细信息
ISBN:
(数字)9781665406734
ISBN:
(纸本)9781665406741
This paper presents a hierarchical distributed optimization algorithm based on quasi-Newton update steps. Separable convex optimization problems are decoupled through dual decomposition and solved in a distributed fashion by coordinating the solutions of the subproblems through dual variables. The proposed algorithm updates the dual variables by approximating the Hessian of the dual function through collected subgradient information, analogously to quasi-Newton methods. As the dual maximization problem is generally nonsmooth, a smooth approximation might show poor performance. To this end cutting planes, analogous to bundle methods, are constructed that take the nonsmoothness of the dual function into account and lead to a better convergence behavior near the optimum. The proposed algorithm is evaluated on a large set of benchmark problems and compared to the subgradient method and to the bundle trust method for nonsmooth optimization.
Operation of robotic manipulators is limited to structured environments and well-defined tasks due to an offline path planning. However, flexible production processes and human-robot collaboration necessitates a real ...
详细信息
ISBN:
(数字)9781665406734
ISBN:
(纸本)9781665406741
Operation of robotic manipulators is limited to structured environments and well-defined tasks due to an offline path planning. However, flexible production processes and human-robot collaboration necessitates a real time path planning to allow for replanning a path in changing environments. In this work, we investigate established planning algorithms for their applicability to dynamic path planning problems. We further compare these methods with our approach based on model predictive control. We consider a single manipulator with six degrees of freedom in static and dynamic environments. We investigate three experimental setups and show the advantages of the proposed MPC-ELS approach over more traditional path planning algorithms in terms of several metrics, such as path-length, execution time or trajectory smoothness. In addition, we propose a scheduling algorithm for object allocation to determine an optimal sequence for pick and place tasks with regard to minimum execution time.
Optical sorting is a key technology for the circular economy and is widely applied in the food, mineral, and recycling industries. Despite its widespread use, one typically resorts to expensive means of adjusting the ...
Optical sorting is a key technology for the circular economy and is widely applied in the food, mineral, and recycling industries. Despite its widespread use, one typically resorts to expensive means of adjusting the accuracy, e.g., by reducing the mass flow or changing mechanical or software parameters, which typically requires manual tuning in a lengthy, iterative process. To circumvent these drawbacks, we propose a new layout for optical sorters along with a controller that allows re-feeding of controlled fractions of the sorted mass flows. To this end, we build a dynamic model of the sorter, analyze its static behavior, and show how material recirculation affects the sorting accuracy. Furthermore, we build a model predictive controller (MPC) employing the model and evaluate the closed-loop sorting system using a coupled discrete element–computational fluid dynamics (DEM–CFD) simulation, demonstrating improved accuracy.
暂无评论