Umeå University's logo

Forestry machines are heavy vehicles performing complex manipulation tasks in unstructured production forest environments. Together with the complex dynamics of the on-board hydraulically actuated cranes, the rough forest terrains have posed a particular challenge in forestry automation. In this study, the feasibility of applying reinforcement learning control to forestry crane manipulators is investigated in a simulated environment. Our results show that it is possible to learn successful actuator-space control policies for energy efficient log grasping by invoking a simple curriculum in a deep reinforcement learning setup. Given the pose of the selected logs, our best control policy reaches a grasping success rate of 97%. Including an energy-optimization goal in the reward function, the energy consumption is significantly reduced compared to control policies learned without incentive for energy optimization, while the increase in cycle time is marginal. The energy-optimization effects can be observed in the overall smoother motion and acceleration profiles during crane manipulation. 

We report on systematic and high-precision measurements of the evolution of electron beams in a laser-wakefield accelerator (LWFA). Utilizing shock-front injection, a technique providing stable, tunable and high-quality electron bunches, acceleration and deceleration of few-MeV quasimonoenergetic beams were measured with cutting-edge technology sub-5-fs and 8-fs laser pulses. We explain the observations with dephasing, an effect that fundamentally limits the performance of LWFAs. Typical density dependent electron energy evolution with 57–300  μm dephasing length and 6–20 MeV peak energy was observed and is well described by a parabolic fit. This is a promising electron source for time-resolved few-fs electron diffraction.

We create partially annotated datasets from field measurements for developing models and algorithms for perception and control of forest machines using artificial intelligence, simulation, and experiments on physical testbeds.  The datasets, algorithms, and trained models for object identification, 3D perception, and motion planning and control will be made publicly available through data and code-sharing repositories.

The data is recorded using forest machines and other equipment with suitable sensors operating in the forest environment. The data include the machine and crane tip position at high resolution, and event time logs (StanForD) while the vehicle operates in high-resolution laser-scanned forest areas.  For annotation, the plan is to use both CAN-bus data and audiovisual data from operators that are willing to participate in the research. Also, by fusing visual perception with operator tree characteristics input or decision, we aim to develop a method for auto-annotation, facilitating a rapid increase in labeled training data for computer vision. In other activities, images of tree plants and bark are collected.

Research questions include, how to automate the process of creating annotated datasets and train models for identifying and positioning forestry objects, such as plants, tree species, logs, terrain obstacles, and do 3D reconstruction for motion planning and control? How large and varied datasets are required for the models to handle the variability in forests, weather, light conditions, etc.? Would additional synthetic data increase model inference accuracy?

In part we focus on forwarders traversing terrain, avoiding obstacles, and loading or unloading logs, with consideration for efficiency, safety, and environmental impact. We explore how to auto-generate and calibrate forestry machine simulators and automation scenario descriptions using the data recorded in the field. The demonstrated automation solutions serve as proofs-of-concept and references, important for developing commercial prototypes and for understanding what future research should focus on.

Fully autonomous forwarding is a challenge, with more imminent scenarios including operator assistance, remote-controlled machines, and semi-autonomous functions. We present several subsystems for autonomous forwarding, developed using machine learning and physics simulation,

- trafficability analysis and path planning,

- autonomous driving,

- identification of logs and high quality grasp poses, and

- crane control from snapshot camera data.

Forwarding is an energy demanding process, and repeated passages with heavy equipment can damage the soil. To avoid damage and ensure efficient use of energy, it is important with a good path planning, adapted speed, and efficient loading and unloading of logs. The collection and availability of large amounts of data is increasing in the field of forestry, opening up for autonomous solutions and efficiency improvements. This is a difficult problem though, as the forest terrain is rough, and as weather, season, obstructions, and wear present challenges in collecting and interpreting sensor-data.

Our proposed subsystems assume access to pre-scanned, high-resolution elevation maps and snapshots of log piles, captured in between crane cycles by an onboard camera. By utilizing snapshots instead of a continuous image stream in the loading task, we separate image segmentation from crane control. This removes any coupling to specific vehicle models, and greatly increases the limit on computational resources and time for the challenge of image segmentation. Log piles are normally static except at the grasp moments and given good enough grasp poses, this lack of information is not necessarily a problem.

We show how snapshot image data can be used when deploying a Reinforcement Learning agent to control the crane to grasp logs in challenging piles. Given pile RGB-D images, our grasp detection model identifies high quality grasp poses, allowing for multiple logs to be loaded in each crane cycle. Further, we show that our model is able to learn to avoid obstructions in the environment such as tree stumps or boulders. We discuss the possibility of using our model to optimize the loading task over a sequence of grasps.

Finally, we discuss how the solutions can be combined in a multi-agent forwarding system with or without a human operator in-the loop.

We review common extensions of particle-in-cell (PIC) schemes which account for strong field phenomena in laser-plasma interactions. After describing the physical processes of interest and their numerical implementation, we provide solutions for several associated methodological and algorithmic problems. We propose a modified event generator that precisely models the entire spectrum of incoherent particle emission without any low-energy cutoff, and which imposes close to the weakest possible demands on the numerical time step. Based on this, we also develop an adaptive event generator that subdivides the time step for locally resolving QED events, allowing for efficient simulation of cascades. Further, we present a unified technical interface for including the processes of interest in different PIC implementations. Two PIC codes which support this interface, PICADOR and ELMIS, are also briefly reviewed.

The validation of a theory is commonly based on appealing to clearly distinguishable and describable features in properly reduced experimental data, while the use of ab-initio simulation for interpreting experimental data typically requires complete knowledge about initial conditions and parameters. We here apply the methodology of using machine learning for overcoming these natural limitations. We outline some basic universal ideas and show how we can use them to resolve long-standing theoretical and experimental difficulties in the problem of high-intensity laser-plasma interactions. In particular we show how an artificial neural network can "read" features imprinted in laser-plasma harmonic spectra that are currently analysed with spectral interferometry.

A scalar Wigner distribution function for describing polarized light is proposed in analogy with the treatment of spin variables in quantum kinetic theory. The formalism is applied to the propagation of circularly polarized light in nonlinear Kerr media, and an extended phase-space evolution equation is derived along with invariant quantities. The formalism is additionally used to analyze the modulational instability. (C) 2013 Optical Society of America

We consider the emission spectrum of high-energy electrons in an intense laser field. At high intensities (a0∼200) we find that the QED theory predicts a narrower angular spread of emissions than the classical theory. This is due to the classical theory overestimating the energy loss of the particles, resulting in them becoming more susceptible to reflection in the laser pulse.

We consider the Thomson and Compton scattering of high-energy electrons n an intense laser pulse. Our simulations show that energy losses due o radiation reaction cause the emitted radiation to be spread over a roader angular range than the case without these losses included. We xplain this in terms of the effect of these energy losses on the article dynamics. Finally, at ultra-high intensities, i.e. fields with dimensionless parameter a(0)similar to 200, the energy of the ission pectrum is significantly reduced by radiation reaction and also the lassical and QED results begin to differ. This is found to be due to he classical theory overestimating the energy loss of the electrons. uch findings are relevant to radiation source development involving e ext generation of high-intensity laser facilities.

Forwarder trafficability is a function of terrain and vehicle properties. Predicting trafficability is vital for energy efficient planning- and operator-assisting systems, as well as for remote and autonomous driving. Inaccurate or insufficient information can lead to inefficient paths, excessive fuel usage, equipment wear, and soil damages. Training trafficability models require data in a quantity hard to collect solely from in-field experiments, especially considering the need for data from situations ranging from very easy to non-traversable.

To circumvent this problem, we perform in-field system identification for a forwarder in the Nordic cut-to-length system, to obtain a calibrated multi-body dynamics simulation model traversing firm but potentially rough and blocky terrain. By letting the real-world forward derdrive in very difficult terrain, the model is able to reflect a wide range of real conditions. The model is used in simulations, where collecting large amounts of data from a variety of situations is easy, cheap, and hazard free. Using this data, a deep neural network is trained to predict trafficability in terms of attainable driving speed, energy consumption, and machine wear.

The resulting predictor model uses laser scanned terrains to efficiently produce trafficability measures with high fidelity and accuracy, e.g., depending on the vehicle’s precise location, speed, heading, and weight. Trafficability on wet and weak soil is not addressed in this work. The predictor model is machine specific, but general enough for practical application in diverse terrain conditions. Our emphasis on energy consumption enables elaborate calculations of emissions, profoundly contributing to sustainable forest operations. Apart from the benefits from reduced emissions, the model can also be used to optimize extraction trail routing, which is a major contributor to the total extraction cost. Rough terrain trafficability is only part of an optima loute, but it has been neglected in previous research. We see big potential in combining our predictor model with existing route optimization methods to achieve a more complete result. By creating an open library of annotated machine data and code for preparing input terrain-data and running the trafficability model, we enable adoption of the results by others and application in existing and new software.

Traversability is of major importance in forestry, where heavy vehicles, weighing up to 40 tons whenfully loaded, traverse rough and sometimes soft terrain. Forest remote sensing is becoming available atresolutions where surface roughness and slope can be determined at length-scales smaller than the forestmachines. Using 3D multibody dynamics simulation of a forest machine driving in virtual terrain replications, the interaction can be captured in great detail. The observed traversability is then automaticallya function of the vehicle geometry, dynamics, and of the local terrain topography relative to heading. Weexpress traversability with three complementary measures: i) the ability to traverse the terrain at a target speed, ii) energy consumption, and iii) machine body acceleration. For high traversability, the lattertwo should be as small as possible while the first measure is at maximum. The simulations are, however,too slow for systematically probing the traversability over large areas. Instead, a deep neural networkis trained to predict the traversability measures from the local heightmap and target speed. The trainingdata comes from simulations of an articulated vehicle with wheeled bogie suspensions driving over procedurally generated terrains while observing the dynamics and local terrain topology. We evaluate themodel on laser-scanned forest terrains, previously unseen by the model. The model predicts traversability with an accuracy of 90% on terrains with 0.25 m resolution and it is 3000 times faster than the groundtruth realtime simulation and trivially parallelizable, making it well suited for traversability analysis andoptimal route planning over large areas. The trained model depends on the vehicle heading, target speed,and detailed features in the topography that a model based only on local slope and roughness cannotcapture. We explore traversability statistics over large areas of laser-scanned terrains and discuss howthe model can be used as a complement or in place of the currently used terrain classification scheme.

The discrete element method (DEM) is a versatile but computationally intensive method for granular dynamics simulation. We investigate the possibility of accelerating DEM simulations using graph neural networks (GNN), which automatically support variable connectivity between particles. This approach was recently found promising for particle-based simulation of complex fluids [1]. We start from a time-implicit, or nonsmooth, DEM [2], where the computational bottleneck is the process of solving a mixed linear complementarity problem (MLCP) to obtain the contact forces and particle velocity update. This solve step is substituted by a GNN, trained to predict the MLCP solution. Following [1], we employ an encoder-process-decoder structure for the GNN. The particle and connectivity data is encoded in an input graph with particle mass, external force, and previous velocity as node attributes, and contact overlap, normal, and tangent vectors as edge attributes. The sought solution is represented in the output graph with the updated particle velocities as node attributes and the contact forces as edge attributes. In the intermediate processing step, the input graph is converted to a latent graph, which is then advanced with a fixed number of message passing steps involving a multilayer perceptron neural network for updating the edge and node values. The output graph, with the approximate solution to the MLCP, is finally computed by decoding the last processed latent graph.

Both a supervised and unsupervised method are tested for training the network on granular simulation of particles in a rotating or static drum. AGX Dynamics [3] is used for running the simulations, and Pytorch [4] in combination with the Deep Graph Library [5] for the learning. The supervised model learns from ground truth MLCP solutions, computed using a projected Gauss-Seidel (PGS) solver, sampled from 1200 simulations involving 50-150 particles. The unsupervised model learns to minimize a loss function derived from the MLCP residual function using particle configurations extracted from the same simulations but ignoring the approximate solution from the PGS solver. The simulation samples are split into training data (80%), validation data (10%), and test data (10%). Network hyperparameter optimization is performed. The supervised GNN solver reaches an error level of 1% for the contact forces and 0.01% on the particle velocities for a static drum. For a rotating drum, the respective errors are 10% and 1%. The unsupervised GNN solver reaches 1% velocity errors, 5% normal forces errors, but it has significant problems with predicting the friction forces. The latter is presumably because of the discontinuous loss function that follows from the Coulomb friction law and therefore we explore regularization of it. Finally, we discuss the potential scalability and performance for large particle systems.

Systems for transport and processing of granular media are challenging to analyse, operate and optimise. In the mining and mineral processing industries, these systems are chains of processes with a complex interplay among the equipment, control and processed material. The material properties have natural variations that are usually only known at certain locations. Therefore, we explored a material-oriented approach to digital twins with a particle representation of the granular media. In digital form, the material is treated as pseudo-particles, each representing a large collection of real particles of various sizes, shapes and mineral properties. Movements and changes in the state of the material are determined by the combined data from control systems, sensors, vehicle telematics and simulation models at locations where no real sensors could see. The particle-based representation enables material tracking along the chain of processes. Each digital particle can act as a carrier of observational data generated by the equipment as it interacts with the real material. This make it possible to better learn the material properties from process observations and to predict the effect on downstream processes. We tested the technique on a mining simulator and demonstrated the analysis that can be performed using data from cross-system material tracking.

The discrete element method (DEM) is a versatile but computationally intense method for simulation of granular materials. It is therefore rarely used in applications that require realtime performance, e.g, interactive simulaions with a human operator or hardware in the loop.

We investigate the use of reduced order modeling for achieving realtime performance in coupled discrete element and rigid multibody simulations. First, a large data set is produced from a series of simulations that cover a selected state-space. The particle data is coarse-grained into discrete field variables, representing mass density, velocity, strain and stress. A reduced order representation of the state-space is identified. Different methods for predicting the fields are explored, given certain observations and assumptions about the state of the simulation e.g., motion of boundaries, rigid bodies or control signals. The particle positions and velocities can then be advanced in time using the predicted fields plus a model for particle diffusion [4] and a local incompressibility constraint [1]. The resulting method can be seen as an extension to the one in [5], by extending the reduced space from rigid body motion of particle aggregates to a low-dimensional space of flow fields [2, 3].

The precision and computational performance of the reduced order simulation method is analyzed on simple test systems, including silo flow and a blade cutting a granular bed. Finally, coupled simulation of an articulated rigid multibody system and a reduced order granular system is demonstrated.

REFERENCES

[1] K. Bodin, C. Lacoursiere, and M. Servin. Constraint fluids. IEEE Transactions on Visualization and Computer Graphics, 18(3):516–526, 2012.

[2] F. Boukouvala, Y. Gao, F. Muzzio, and M. Ierapetritou. Reduced-order discrete element method modeling. Chemical Engineering Science, 95:12–26, 2013.

[3] A. Rogers and M. Ierapetritou. Discrete element reduced-order modeling of dynamicparticulate systems. AIChE Journal, 60:3184–94, 2014.

[4] P. Salamon, D. Fernandez-Garcia, and J. G ` omez-Hern ´ andez. A review and numerical assessment of the random walk particle tracking method. Journal of Contaminant Hydrology, 87(3):277 – 305, 2006.

[5] M. Servin and D. Wang. Adaptive model reduction for nonsmooth discrete element simulation. Computational Particle Mechanics, 3(1):107–121, 2016.

We model the emission of high energy photons due to relativistic charged particle motion in intense laser-plasma interactions. This is done within a particle-in-cell code, for which high frequency radiation normally cannot be resolved due to finite time steps and grid size. A simple expression for the synchrotron radiation spectra is used together with a Monte-Carlo method for the emittance. We extend previous work by allowing for arbitrary fields, considering the particles to be in instantaneous circular motion due to an effective magnetic field. Furthermore, we implement noise reduction techniques and present validity estimates of the method. Finally, we perform a rigorous comparison to the mechanism of radiation reaction, and find the emitted energy to be in excellent agreement with the losses calculated using radiation reaction. (C) 2015 AIP Publishing LLC.

We investigate the use of reduced-order modelling to run discrete element simulations at higher speeds. Taking a data-drivenapproach, we run many offline simulations in advance and train a model to predict the velocity field from the mass distributionand system control signals. Rapid model inference of particle velocities replaces the intense process of computing contactforces and velocity updates. In coupled DEM and multibody system simulation, the predictor model can be trained to outputthe interfacial reaction forces as well. An adaptive model order reduction technique is investigated, decomposing the mediain domains of solid, liquid, and gaseous state. The model reduction is applied to solid and liquid domains where the particlemotion is strongly correlated with the mean flow, while resolved DEM is used for gaseous domains. Using a ridge regressionpredictor, the performance is tested on simulations of a pile discharge and bulldozing. The measured accuracy is about 90%and 65%, respectively, and the speed-up range between 10 and 60. 

We explore multi-log grasping using reinforcement learning and virtual visual servoing for automated forwarding in a simulated environment. Automation of forest processes is a major challenge, and many techniques regarding robot control pose different challenges due to the unstructured and harsh outdoor environment. Grasping multiple logs involves various problems of dynamics and path planning, where understanding the interaction between the grapple, logs, terrain, and obstacles requires visual information. To address these challenges, we separate image segmentation from crane control and utilise a virtual camera to provide an image stream from reconstructed 3D data. We use Cartesian control to simplify domain transfer to real-world applications. Because log piles are static, visual servoing using a 3D reconstruction of the pile and its surroundings is equivalent to using real camera data until the point of grasping. This relaxes the limits on computational resources and time for the challenge of image segmentation, and allows for data collection in situations where the log piles are not occluded. The disadvantage is the lack of information during grasping. We demonstrate that this problem is manageable and present an agent that is 95% successful in picking one or several logs from challenging piles of 2–5 logs.

We present a method that uses high-resolution topography data of rough terrain, and ground vehicle simulation, to predict traversability. Traversability is expressed as three independent measures: the ability to traverse the terrain at a target speed, energy consumption, and acceleration. The measures are continuous and reflect different objectives for planning that go beyond binary classification. A deep neural network is trained to predict the traversability measures from the local heightmap and target speed. To produce training data, we use an articulated vehicle with wheeled bogie suspensions and procedurally generated terrains. We evaluate the model on laser-scanned forest terrains, previously unseen by the model. The model predicts traversability with an accuracy of 90%. Predictions rely on features from the high-dimensional terrain data that surpass local roughness and slope relative to the heading. Correlations show that the three traversability measures are complementary to each other. With an inference speed 3000 times faster than the ground truth simulation and trivially parallelizable, the model is well suited for traversability analysis and optimal path planning over large areas.

The theory for nonlinear three-wave interaction in magnetized plasmas is reconsidered using quantum hydrodynamics. The general coupling coefficients are calculated for the generalized Bohm de Broglie term. It is found that the Manley-Rowe relations are fulfilled only if the form of the particle dispersive term coincides with the standard expression. The implications of our results are discussed.

We explore sim-to-real transfer of deep reinforcement learning controllers for a heavy vehicle with active suspensions designed for traversing rough terrain. While related research primarily focuses on lightweight robots with electric motors and fast actuation, this study uses a forestry vehicle with a complex hydraulic driveline and slow actuation. We simulate the vehicle using multibody dynamics and apply system identification to find an appropriate set of simulation parameters. We then train policies in simulation using various techniques to mitigate the sim-to-real gap, including domain randomization, action delays, and a reward penalty to encourage smooth control. In reality, the policies trained with action delays and a penalty for erratic actions perform at nearly the same level as in simulation. In experiments on level ground, the motion trajectories closely overlap when turning to either side, as well as in a route tracking scenario. When faced with a ramp that requires active use of the suspensions, the simulated and real motions are in close alignment. This shows that the actuator model together with system identification yields a sufficiently accurate model of the actuators. We observe that policies trained without the additional action penalty exhibit fast switching or bang-bang control. These present smooth motions and high performance in simulation but transfer poorly to reality. We find that policies make marginal use of the local height map for perception, showing no indications of look-ahead planning. However, the strong transfer capabilities entail that further development concerning perception and performance can be largely confined to simulation. 

We explore sim-to-real transfer of deep reinforcement learning controllers for a heavy vehicle with active suspensions designed for traversing rough terrain. While related research primarily focuses on lightweight robots with electric motors and fast actuation, this study uses a forestry vehicle with a complex hydraulic driveline and slow actuation. We simulate the vehicle using multibody dynamics and apply system identification to find an appropriate set of simulation parameters. We then train policies in simulation using various techniques to mitigate the sim-to-real gap, including domain randomization, action delays, and a reward penalty to encourage smooth control. In reality, the policies trained with action delays and a penalty for erratic actions perform nearly at the same level as in simulation. In experiments on level ground, the motion trajectories closely overlap when turning to either side, as well as in a route tracking scenario. When faced with a ramp that requires active use of the suspensions, the simulated and real motions are in close alignment. This shows that the actuator model together with system identification yields a sufficiently accurate model of the actuators. We observe that policies trained without the additional action penalty exhibit fast switching or bang–bang control. These present smooth motions and high performance in simulation but transfer poorly to reality. We find that policies make marginal use of the local height map for perception, showing no indications of predictive planning. However, the strong transfer capabilities entail that further development concerning perception and performance can be largely confined to simulation.

We explore the potential to control terrain vehicles using deep reinforcement in scenarios where human operators and traditional control methods are inadequate. This letter presents a controller that perceives, plans, and successfully controls a 16-tonne forestry vehicle with two frame articulation joints, six wheels, and their actively articulated suspensions to traverse rough terrain. The carefully shaped reward signal promotes safe, environmental, and efficient driving, which leads to the emergence of unprecedented driving skills. We test learned skills in a virtual environment, including terrains reconstructed from high-density laser scans of forest sites. The controller displays the ability to handle obstructing obstacles, slopes up to 27 degrees, and a variety of natural terrains, all with limited wheel slip, smooth, and upright traversal with intelligent use of the active suspensions. The results confirm that deep reinforcement learning has the potential to enhance control of vehicles with complex dynamics and high-dimensional observation data compared to human operators or traditional control methods, especially in rough terrain.

Automating the loaded and unloaded driving of a forwarder has the potential to reduce operational costsup to 10% in cut-to-length logging, but remains a challening and unsolved task. The complex interaction between the vehicle and terrain requires the controller to percieve its surroundings and thestate of the vehicle to plan for traversal. Because the state space is high dimensional and the systemdynamics cannot be formulated in closed form or easily approximated, traditional control methods areinadequate. Under these conditions, where learning to act in the environement is easier than learning thesystem dynamics, model free reinforcement learning is a promising option. We use deep reinforcementlearning for control of a 16-tonne forwarder with actively articulated suspensions. To efficiently gathergeneralizable experience, the control policies were safetly trained in simulation while varying severaldomain parameters. Each policy is trained during what correponds to roughly one month of real time.In simulation, the controller shows the ability to traverse rough terrains reconstruced from high-densitylaser scans and handles slopes up to 27◦. To compare the simulated to real performance we transfer thecontrol policies to the physical vehicle. Our results provide insight on how to improve policy transfer toheavy and expensive forest machines.