Embedded real-time multi-baseline stereo

Honegger, Dominik; Sattler, Torsten; Pollefeys, Marc

doi:10.1109/icra.2017.7989615

Cited by 30 publications

(34 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To perceive the bush to be trimmed, the robot uses a stereo‐camera system above the end‐effector (see Figure ), developed at ETH Zürich, Switzerland (Honegger, Sattler, & Pollefeys, ). The two cameras had a diagonal FOV of

6 8^{\circ}

.…”

Section: Field Test Of the Coverage Algorithmmentioning

confidence: 99%

Coverage trajectory planning for a bush trimming robot arm

Kaljaca

Vroegindeweij

Henten

2019

Journal of Field Robotics

View full text Add to dashboard Cite

A novel motion planning algorithm for robotic bush trimming is presented. The algorithm is based on an optimal route search over a graph. Differently from other works in robotic surface coverage, it entails both accuracy in the surface sweeping task and smoothness in the motion of the robot arm. The proposed method requires the selection of a custom objective function in the joint space for optimal node traversal scheduling, as well as a kinematically constrained time interpolation. The algorithm was tested in simulation using a model of the Jaco arm and three target bush shapes. Analysis of the simulated motions showed how, differently from classical coverage techniques, the proposed algorithm is able to ensure high tool positioning accuracy while avoiding excessive arm motion jerkiness. It was reported that forbidding manipulation posture changes during the cutting phase of the motion is a key element for task accuracy, leading to a decrease of the tool positioning error up to 90%. Furthermore, the algorithm was validated in a real‐world trimming scenario with boxwood bushes. A target of 20 mm accuracy was proposed for a trimming result to be considered successful. Results showed that on average 82% of the bush surface was affected by trimming, and 51% of the trimmed surface was cut within the desired level of accuracy. Despite the fact that the trimming accuracy turned out to be lower than the stated requirements, it was found out this was mainly a consequence of the inaccurate, early stage vision system employed to compute the target trimming surface. By contrast, the trimming motion planning algorithm generated trajectories that smoothly followed their input target and allowed effective branch cutting.

show abstract

6 8^{\circ}

.…”

Section: Field Test Of the Coverage Algorithmmentioning

confidence: 99%

Coverage trajectory planning for a bush trimming robot arm

Kaljaca

Vroegindeweij

Henten

2019

Journal of Field Robotics

View full text Add to dashboard Cite

show abstract

“…The network is implemented using TensorFlow [19] and trained & tested using an Intel Core i7-7820HK processor (quad-core, 8MB cache, up to 4.4GHZ) and Nvidia Geforce GTX 1080Ti. First, an ablation study with initial disparity inputs ( [3,4]) is conducted using a synthetic garden dataset to analyze the influence of each factor in the energy function. Secondly, a real test on the Kitti2015 dataset [20] is done with two initial inputs ( [2,5]).…”

Section: Methodsmentioning

confidence: 99%

“…The reason using a synthetic dataset is that the real dataset (eg: Kitti2015) does not have dense ground truth, which will influence the evaluation of the network. Dispnet [3] and FPGA-stereo [4] were used as inputs. The authors of [3,4] helped us get the best performance on the dataset as the input to the network.…”

Section: Ablation Studymentioning

confidence: 99%

See 1 more Smart Citation

UDFNET: Unsupervised Disparity Fusion with Adversarial Networks

Fisher

2019

2019 IEEE International Conference on Image Processing (ICIP)

View full text Add to dashboard Cite

Existing disparity fusion methods based on deep learning achieve state-of-the-art performance, but they require ground truth disparity data to train. As far as I know, this is the first time an unsupervised disparity fusion not using ground truth disparity data has been proposed. In this paper, a mathematical model for disparity fusion is proposed to guide an adversarial network to train effectively without ground truth disparity data. The initial disparity maps are inputted from the left view along with auxiliary information (gradient, left & right intensity image) into the refiner and the refiner is trained to output the refined disparity map registered on the left view. The refined left disparity map and left intensity image are used to reconstruct a fake right intensity image. Finally, the fake and real right intensity images (from the right stereo vision camera) are fed into the discriminator. In the model, the refiner is trained to output a refined disparity value close to the weighted sum of the disparity inputs for global initialisation. Then, three refinement principles are adopted to refine the results further. (1) The reconstructed intensity error between the fake and real right intensity image is minimised.(2) The similarities between the fake and real right image in different receptive fields are maximised.(3) The refined disparity map is smoothed based on the corresponding intensity image. The adversarial networks' architectures are effective for the fusion task. The fusion time using the proposed network is small. The network can achieve 90 fps using Nvidia Geforce GTX 1080Ti on the Kitti2015 dataset when the input resolution is 1242 * 375 (Width * Height) without downsampling and cropping. The accuracy of this work is equal to (or better than) the state-of-the-art supervised methods.

show abstract

“…al. [8] or PIRVS [9] are either unavailable and/or are limited in hardware configuration regarding image sensor, lens, camera baseline and IMU. Furthermore, it is often impossible to add further extensions of those frameworks to enable fusion with other modalities such as LiDAR sensors or external illumination.…”

Section: Introductionmentioning

confidence: 99%

VersaVIS—An Open Versatile Multi-Camera Visual-Inertial Sensor Suite

Tschopp

Riner

Bernreiter

et al. 2020

Sensors

View full text Add to dashboard Cite

Robust and accurate pose estimation is crucial for many applications in mobile robotics. Extending visual Simultaneous Localization and Mapping (SLAM) with other modalities such as an inertial measurement unit (IMU) can boost robustness and accuracy. However, for a tight sensor fusion, accurate time synchronization of the sensors is often crucial. Changing exposure times, internal sensor filtering, multiple clock sources and unpredictable delays from operation system scheduling and data transfer can make sensor synchronization challenging. In this paper, we present VersaVIS, an Open Versatile Multi-Camera Visual-Inertial Sensor Suite aimed to be an efficient research platform for easy deployment, integration and extension for many mobile robotic applications. VersaVIS provides a complete, open-source hardware, firmware and software bundle to perform time synchronization of multiple cameras with an IMU featuring exposure compensation, host clock translation and independent and stereo camera triggering. The sensor suite supports a wide range of cameras and IMUs to match the requirements of the application. The synchronization accuracy of the framework is evaluated on multiple experiments achieving timing accuracy of less than 1 ms. Furthermore, the applicability and versatility of the sensor suite is demonstrated in multiple applications including visual-inertial SLAM, multi-camera applications, multimodal mapping, reconstruction and object based mapping.

show abstract

Embedded real-time multi-baseline stereo

Cited by 30 publications

References 17 publications

Coverage trajectory planning for a bush trimming robot arm

Coverage trajectory planning for a bush trimming robot arm

UDFNET: Unsupervised Disparity Fusion with Adversarial Networks

VersaVIS—An Open Versatile Multi-Camera Visual-Inertial Sensor Suite

Contact Info

Product

Resources

About