Towards Finding the Optimal Bucket Filling Strategy through Simulation

Filla, Reno; Frank, Bobbie

doi:10.3384/ecp17144402

Cited by 15 publications

(14 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…An overview of the research from the late 1980s to 2009 can be found in [3]. Previous research has focused on computer vision in order to characterize piles [4][5][6], the motion planning of bucket-filling trajectories [4,[7][8][9], and loading control. The control strategies can be divided into trajectory control [8,10] and force-feedback control [11,12].…”

Section: Related Work and Our Contributionmentioning

confidence: 99%

“…Previous research has focused on computer vision in order to characterize piles [4][5][6], the motion planning of bucket-filling trajectories [4,[7][8][9], and loading control. The control strategies can be divided into trajectory control [8,10] and force-feedback control [11,12]. The former aims to move the bucket along a pre-defined path, which is limited to homogeneous media, such as dry sand and gravel.…”

Section: Related Work and Our Contributionmentioning

confidence: 99%

See 1 more Smart Citation

Continuous Control of an Underground Loader Using Deep Reinforcement Learning

et al. 2021

View full text Add to dashboard Cite

The reinforcement learning control of an underground loader was investigated in a simulated environment by using a multi-agent deep neural network approach. At the start of each loading cycle, one agent selects the dig position from a depth camera image of a pile of fragmented rock. A second agent is responsible for continuous control of the vehicle, with the goal of filling the bucket at the selected loading point while avoiding collisions, getting stuck, or losing ground traction. This relies on motion and force sensors, as well as on a camera and lidar. Using a soft actor–critic algorithm, the agents learn policies for efficient bucket filling over many subsequent loading cycles, with a clear ability to adapt to the changing environment. The best results—on average, 75% of the max capacity—were obtained when including a penalty for energy usage in the reward.

show abstract

Section: Related Work and Our Contributionmentioning

confidence: 99%

Section: Related Work and Our Contributionmentioning

confidence: 99%

Continuous Control of an Underground Loader Using Deep Reinforcement Learning

et al. 2021

View full text Add to dashboard Cite

show abstract

“…An overview of the research from the late 1980's to 2009 is found in [3]. Previous research has focused on computer vision for characterizing the pile [4,5,6], motion planning of bucket filling trajectories [7,4,8,9], and of loading control. The control strategies can be divided in trajectory control [10,8] or force-feedback control [11,12].…”

Section: Related Workmentioning

confidence: 99%

“…Previous research has focused on computer vision for characterizing the pile [4,5,6], motion planning of bucket filling trajectories [7,4,8,9], and of loading control. The control strategies can be divided in trajectory control [10,8] or force-feedback control [11,12]. The former aims to move the bucket along a pre-defined path, which is limited to homogeneous media such as dry sand and gravel.…”

Section: Related Workmentioning

confidence: 99%

Continuous control of an underground loader using deep reinforcement learning

Backman¹,

Lindmark²,

Bodin³

et al. 2021

Preprint

View full text Add to dashboard Cite

Reinforcement learning control of an underground loader is investigated in simulated environment, using a multi-agent deep neural network approach. At the start of each loading cycle, one agent selects the dig position from a depth camera image of the pile of fragmented rock. A second agent is responsible for continuous control of the vehicle, with the goal of filling the bucket at the selected loading point, while avoiding collisions, getting stuck, or losing ground traction. It relies on motion and force sensors, as well as on camera and lidar. Using a soft actor-critic algorithm the agents learn policies for efficient bucket filling over many subsequent loading cycles, with clear ability to adapt to the changing environment. The best results, on average 75% of the max capacity, are obtained when including a penalty for energy usage in the reward.

show abstract

“…However, as described in [3] the variation for one individual operator also varies greatly. [4] Operability is the ease with which the user can operate the machine as intended. As the operators are essential to the performance of the machine, operability is important.…”

Section: On Energy Efficient Mobile Hydraulic Systemsmentioning

confidence: 99%

On Energy Efficient Mobile Hydraulic Systems: with Focus on Linear Actuation

Heybroek¹

View full text Add to dashboard Cite

In this dissertation, energy efficient hydraulic systems are studied. The research focuses on solutions for linear actuators in mobile applications, with emphasis on construction machines. Alongside the aspect of energy efficiency, the thesis deals with competing aspects in hydraulic system design found in the development of construction machines. Simulation models and controls for different concepts are developed, taking the whole machine into account. In line with this work, several proof of concept demonstrators are developed. In the thesis three main system topologies are covered:First, pump controlled systems are studied and a novel concept based on an open-circuit pump configuration is conceived. Special consideration is paid to multi-mode capabilities that allow for a broadened operating range and potential downsizing of components. Simulation models and controls are developed and the system is experimentally validated in a wheel loader application.Second, the possibility for energy recuperation in valve-controlled systems is investigated. In such solutions, a hydraulic motor, added to the meter-out port, is used for energy recovery during load lowering and in multi-function operation. Recuperated energy is either used momentarily or stored in a hydraulic accumulator. The proposed solution means an incremental improvement to conventional systems, which is sometimes attractive to machine manufacturers due to fewer uncertainties in reliability, safety and development cost. The energy recovery system is studied on a conceptual level where several alternative systems are proposed and a concept based on a two-machine hydraulic transformer is selected for a deeper control study followed by experimental validation.Third, so-called common pressure rail systems are considered. This technique is well established for rotary drives, at least for the industrial sector. However, in applying this technique to mobile hydraulics, feasible solutions for linear actuators are needed. In this dissertation, two approaches to this problem are presented. The first one focuses on hydraulic pressure transformers and the second one on secondary controlled multi-chamber cylinders. i ii Populärvetenskaplig sammanfattningHuvudtemat för denna avhandling är energieffektivisering av hydraulsystem. Forskningen rör främst linjära rörelsesystem inom mobila tillämpningar, med fokus på anläggningsmaskiner. Avhandlingen berör åtskilliga aspekter av systemdesign, där simulering och styrning av dynamiska system samt helfordonsmodellering är återkommande inslag. Ett led i forskningsstudien har varit att bygga in och utvärdera framtagna koncept genom fullskaliga demonstratorer. Ett flertal nya hydraulsystem har studerats inom tre huvudsakliga inriktningar: En första inriktning är mot så kallade pumpstyrda system, där ett nytt koncept baserat på pumpar anslutna i en öppen krets har utvecklats. Inom detta har smart reglering av ventiler visat sig vara central för att uppnå en hög energieffektivitet över ett brett arbetsområde. Konceptet vali...

show abstract

Towards Finding the Optimal Bucket Filling Strategy through Simulation

Cited by 15 publications

References 6 publications

Continuous Control of an Underground Loader Using Deep Reinforcement Learning

Continuous Control of an Underground Loader Using Deep Reinforcement Learning

Continuous control of an underground loader using deep reinforcement learning

On Energy Efficient Mobile Hydraulic Systems: with Focus on Linear Actuation

Contact Info

Product

Resources

About