117 research outputs found
Design and implementation of a binary classifier for an ultra-light weight perching system for sloped or vertical rough surfaces on Mars.
No full textThesis done by Matko Matic at KU Leuven under supervision of professor Renaud Detry and professor Tinne Tuytelaar
Replication data for: Fracture toughness and auxeticity in disordered metamaterials
No full text
Overview
This dataset contains both raw and processed data acquired from experimental mechanical tests and computational simulations.
The data is organized to characterize and optimize the mechanical behavior of auxetic and fracture-resistant metamaterial structures.
Experimental Data: Uniaxial tensile tests, synchronized high-resolution video recordings, and digital image correlation (DIC)-based analysis.
Simulation Data: Results from stochastic optimization of soft-disk generated metamaterial networks under tensile and compressive loading.
Experimental Data
Directory Structure
250408_TensileTest/ — Stress–strain data in .txt format for non-optimized and optimized samples.
250408_NONOPT#_Video/, 250408_OPT#_Video/ — High-resolution .tif image sequences recorded during mechanical tests.
250523_ElaboratedVideo/ — Processed video data including:
250523_CellTrack/ — Cell deformation tracking for auxeticity quantification.
250523_GanasceTrack/ — Grip displacement tracking for axial strain determination.
250523_TrackSagome/ — Boundary tracking of gauge region for lateral strain computation.
File Naming Convention
Files follow the pattern: 250408_<SAMPLE_ID>_TensileTest.txt
SAMPLE_ID corresponds to:
NONOPT# — Non-optimized specimen
OPT# — Optimized specimen
Structure of Tensile Test Files
Metadata header containing:
Test Method
Sample I.D.
Specimen Number
Data table, semicolon-delimited, including:
Load (N)
Time (s)
Extension (mm)
Stress (MPa)
Strain (%)
Tensile tests were performed using a universal testing machine and synchronized with high-speed video acquisition for post-processing via image-based strain mapping.
Frame-by-Frame Video Data
Files named as: 250408_<SAMPLE_ID>_TensileTest_###.tif
Each file corresponds to a single frame (starting from 001).
High-speed grayscale recordings used for strain field computation, Poisson ratio estimation, and fracture detection.
Simulation Data
Directory Structure
Located under simulation_DATA/ and organized into four loading scenarios:
COMP-M/ — Compression, mass redistribution along M direction.
COMP-N/ — Compression, mass redistribution along N direction.
TENS-M/ — Tension, mass redistribution along M direction.
TENS-N/ — Tension, mass redistribution along N direction.
Sample Structure and Replication
25 network samples generated via soft-disk simulations.
Each sample tested twice in each batch.
Total optimization runs: 4 batches × 25 samples × 2 replicas = 200.
Subdirectory Naming Scheme
YYYYMMDD_sample-<sample-id>_replica-<replica-id>/
Files in Each Optimization Run
nodes_init.dat — Initial nodal coordinates (x, y).
nodes_final.dat — Final nodal coordinates (x, y).
edges_init.dat — Initial edge connectivity.
edges_final.dat — Final edge connectivity.
force_disp_init.dat — Initial force–displacement curve.
force_disp_final.dat — Final force–displacement curve.
opt.txt — Summary of optimization parameters (weight_nu, weight_U, nu_init, nu_final, U_init, U_final).
edge_data_final.dat — Final beam-level data (bending ratio, von Mises stress, GEBC).
Acknowledgements
A.L.H.S.D., R.G. and S.Z. acknowledge support from the ARCHIBIOFOAM project,
which received funding from the European Union’s Horizon Europe research and
innovation programme under grant agreement No 101161052. Views and opinions
expressed are however those of the author(s) only and do not necessarily reflect
those of the European Union or European Innovation Council and SMEs Executive Agency (EISMEA).
Neither the European Union nor the granting authority can be held responsible for them.
</section
Design and implementation of a binary classifier for an ultra-light weight perching system for sloped or vertical rough surfaces on Mars.
No full textThesis done by Matko Matic at KU Leuven under supervision of professor Renaud Detry and professor Tinne Tuytelaar
Learning of Multi-Dimensional, Multi-Modal Features for Robotic Grasping
No full textWhile robots are extensively used in factories, our industry hasn't yet been able to prepare them for working in human environments - for instance in houses or in human-operated factories. The main obstacle to these applications lies in the amplitude of the uncertainty inherent to the environments humans are used to work in, and in the difficulty in programming robots to cope with it. For instance, in robot-oriented environments, robots can expect to find specific tools and objects in specific places. In a human environment, obstacles may force one to find a new way of holding a tool, and new objects appear continuously and need to be dealt with. As it proves difficult to build into robots the knowledge necessary for coping with uncertain environments, the robotics community is turning to the development of agents that acquire this knowledge progressively and that adapt to unexpected events.
This thesis studies the problem of vision-based robotic grasping in uncertain environments. We aim to create an autonomous agent that develops grasping skills from experience, by interacting with objects and with other agents. To this end, we present a 3D object model for autonomous, visuomotor interaction. The model represents grasping strategies along with visual features that predict their applicability. It provides a robot with the ability to compute grasp parameters from visual observations. The agent acquires models interactively by manipulating objects, possibly imitating a teacher. With time, it becomes increasingly efficient at inferring grasps from visual evidence. This behavior relies on (1) a grasp model representing relative object-gripper configurations and their feasibility, and (2) a model of visual object structure, which aligns the grasp model to arbitrary object poses (3D positions and orientations).
The visual model represents object edges or object faces in 3D by probabilistically encoding the spatial distribution of small segments of object edges or the distribution of small patches of object surface. A model is learned from a few segmented 3D scans or stereo images of an object. Monte Carlo simulation provides robust estimates of the object's 3D position and orientation in cluttered scenes.
The grasp model represents the likelihood of success of relative object-gripper configurations. Initial models are acquired from visual cues or by observing a teacher. Models are then refined autonomously by ``playing' with objects and observing the effects of exploratory grasps. After the robot has learned a few object models, learning becomes a combination of cross-object generalization and interactive experience: grasping strategies are generalized across objects that share similar visual substructures; they are then adapted to new objects through autonomous exploration.
The applicability of our model is supported by numerous examples of pose estimates in cluttered scenes, and by a robot platform that shows increasing grasping capabilities as it explores its environment
Neuromorphic tactile sensor array based on fiber Bragg gratings to encode object qualities
No full textEmulating the sense of touch is fundamental to endow robotic systems with perception abilities. This work presents an unprecedented mechanoreceptor-like neuromorphic tactile sensor implemented with fiber optic sensing technologies. A robotic gripper was sensorized using soft and flexible tactile sensors based on Fiber Bragg Grating (FBG) transducers and a neuro-bio-inspired model to extract tactile features. The FBGs connected to the neuron model emulated biological mechanoreceptors in encoding tactile information by means of spikes. This conversion of inflowing tactile information into event-based spikes has an advantage of reduced bandwidth requirements to allow communication between sensing and computational subsystems of robots. The outputs of the sensor were converted into spiking on-off events by means of an architecture implemented in a Field Programmable Gate Array (FPGA) and applied to robotic manipulation tasks to evaluate the effectiveness of such information encoding strategy. Different tasks were performed with the objective to grant fine manipulation abilities using the features extracted from the grasped objects (i.e., size and hardness). This is envisioned to be a futuristic sensor technology combining two promising technologies: optical and neuromorphic sensing
Improving Generalization of Synthetically Trained Sonar Image Descriptors for Underwater Place Recognition
No full textAutonomous navigation in underwater environments presents challenges due to
factors such as light absorption and water turbidity, limiting the
effectiveness of optical sensors. Sonar systems are commonly used for
perception in underwater operations as they are unaffected by these
limitations. Traditional computer vision algorithms are less effective when
applied to sonar-generated acoustic images, while convolutional neural networks
(CNNs) typically require large amounts of labeled training data that are often
unavailable or difficult to acquire. To this end, we propose a novel compact
deep sonar descriptor pipeline that can generalize to real scenarios while
being trained exclusively on synthetic data. Our architecture is based on a
ResNet18 back-end and a properly parameterized random Gaussian projection
layer, whereas input sonar data is enhanced with standard ad-hoc
normalization/prefiltering techniques. A customized synthetic data generation
procedure is also presented. The proposed method has been evaluated extensively
using both synthetic and publicly available real data, demonstrating its
effectiveness compared to state-of-the-art methods.Comment: This paper has been accepted for publication at the 14th
International Conference on Computer Vision Systems (ICVS 2023
Grasp Generalization Via Predictive Parts
No full textObjects can be grasped in various ways. Depending on the scenario (object shape, grip-per/hand, task objectives,...), grasps that differ in object-relative pose may greatly differ in their utility. In previous work [1, 2] we introduced the concept of grasp densities that represent, for a given scenario, the distribution of successful, object-relative gripper poses pX|O=success(x), (1) a probability density function over the object-relative gripper poses, represented as a random variable X in SE(3) = SO(3) × R3, given that the grasp outcome O is success. Using maximum-likelihood reasoning, suitable grasps can be chosen according to this den-sity, even if parts of the density are pruned by influences such as obstructions, kinematic constraints, etc. Grasp densities are designed to be learned empirically. In principle, a robot attemps to grasp the object a large number of times using a wide variety of object-relative gripper poses. Each successful grasp constitutes a data point drawn from the underlying grasp density (1). In practice, for reasons of efficiency, attempted grasps should be chosen in an informed manner [1]. For resampling and inference, samples are turned into a continuou
Continuous Surface-point Distributions for 3D Object Pose Estimation and Recognition
No full textpeer reviewe
- …
