Playing Soccer Without Colors in the SPL: A Convolutional Neural Network Approach

Leiva, Francisco; Cruz, Nicolás; Bugueño, Ignacio; Ruiz‐del‐Solar, Javier

doi:10.1007/978-3-030-27544-0_10

Cited by 11 publications

(9 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Speck et al [5] proposed an approach that uses neural networks for real-time classification and detection on a humanoid nao robot. Apart from deep learning methods, there are approaches that depends on colour classification, that works with the idea that the entire soccer environment is green and the objects can be recognized by finding the gaps as proposed by Hoffmann et al [6], Lenser and Veloso [7], Leiva et al [8], Dijk and Scheunemann [9] proposed a neural network based semantic segmentation that combines the robot vision techniques on grayscale images and the convolutional neural network (CNN) classification. The object detection and tracking on the robosoccer environment is a complex problem as it depends on various factors like shadows, reflections, obstructions, vanishing points, high activity points and there are main challenges like occlusions, speed, multiple scales and limited data.…”

Section: The Development Trend In Robocup On Object Detection and Tra...mentioning

confidence: 99%

Object detection on robosoccer environment using convolution neural network

Steffi¹,

Mehta²,

Venkatesh

2022

IJEECS

View full text Add to dashboard Cite

Robots with autonomous capabilities depend on vision capabilities to detect and interact with objects and their environment. In the field of robotic research, one of the focus areas is the robosoccer platform that is being used to implement and test new ideas and findings on computer vision and decision making. In this article, an efficient real-time object detection algorithm is employed in a robosoccer simulation environment by deploying a convolution neural network and Kalman filter based tracking algorithms. This study's objective is to classify nao, ball, and the goalpost as well as to validate nao and ball tracking without human intervention from initial frame to last frame. In comparison with the existing methods, the proposed method is robust and fast in identifying three classes namely nao, ball, and goalpost with a speed of 1.67 FPS and a mAP of 95.18%. By implementing this approach, soccer playing robots can make appropriate decisions during game play.

show abstract

Section: The Development Trend In Robocup On Object Detection and Tra...mentioning

confidence: 99%

Object detection on robosoccer environment using convolution neural network

Steffi¹,

Mehta²,

Venkatesh

2022

IJEECS

View full text Add to dashboard Cite

show abstract

“…The training time of this classifier was about 10 hours that is an issue for on-site training during the competitions. Recently several neural network based classifiers are presented for classic ball detection in standard platform league [11][12][13]. In [11] ball candidates are examined using black pentagons of the ball.…”

Section: Related Workmentioning

confidence: 99%

“…The topology and configurations of the proposed network are optimized using a genetic algorithm. A more general candidate generation is suggested by [12]. For each interested pixel, a Difference of Gaussian filter with a kernel size that is proportional to the expected ball radius at the location of the pixel is applied.…”

Section: Related Workmentioning

confidence: 99%

A Real-Time Ball Detection Approach Using Convolutional Neural Networks

Teimouri

Delavaran

Rezaei

2019

RoboCup 2019: Robot World Cup XXIII

View full text Add to dashboard Cite

Ball detection is one of the most important tasks in the context of soccer-playing robots. The ball is a small moving object which can be blurred and occluded in many situations. Several neural network based methods with different architectures are proposed to deal with the ball detection. However, they are either neglecting to consider the computationally low resources of humanoid robots or highly depend on manually-tuned heuristic methods to extract the ball candidates. In this paper, we propose a new ball detection method for low-cost humanoid robots that can detect most soccer balls with a high accuracy rate of up to 97.17%. The proposed method is divided into two steps. First, some coarse regions that may contain a full ball are extracted using an iterative method employing an efficient integral image based feature. Then they are fed to a lightweight convolutional neural network to finalize the bounding box of a ball. We have evaluated the proposed approach using a comprehensive dataset and the experimental results show the efficiency of our method.

show abstract

“…However, in light of RoboCup's 2050 goal, teams have seen the introduction of natural lighting conditions (exposure to sunlight), white goal posts with white backgrounds and FIFA balls with a variety of colours. Colour segmentation based techniques fail to perform in these challenging scenarios and has mostly pushed the competition towards implementing a variety of neural network approaches [12] [25].…”

Section: Related Workmentioning

confidence: 99%

xYOLO: A Model For Real-Time Object Detection In Humanoid Soccer On Low-End Hardware

Barry

Shah

Keijsers

et al. 2019

2019 International Conference on Image and Vision Computing New Zealand (IVCNZ)

View full text Add to dashboard Cite

With the emergence of onboard vision processing for areas such as the internet of things (IoT), edge computing and autonomous robots, there is increasing demand for computationally efficient convolutional neural network (CNN) models to perform real-time object detection on resource constraints hardware devices. Tiny-YOLO is generally considered as one of the faster object detectors for low-end devices and is the basis for our work. Our experiments on this network have shown that Tiny-YOLO can achieve 0.14 frames per second (FPS) on the Raspberry Pi 3 B, which is too slow for soccer playing autonomous humanoid robots detecting goal and ball objects. In this paper we propose an adaptation to the YOLO CNN model named xYOLO, that can achieve object detection at a speed of 9.66 FPS on the Raspberry Pi 3 B. This is achieved by trading an acceptable amount of accuracy, making the network approximately 70 times faster than Tiny-YOLO. Greater inference speed-ups were also achieved on a desktop CPU and GPU. Additionally we contribute an annotated Darknet dataset for goal and ball detection.

show abstract

Playing Soccer Without Colors in the SPL: A Convolutional Neural Network Approach

Cited by 11 publications

References 14 publications

Object detection on robosoccer environment using convolution neural network

Object detection on robosoccer environment using convolution neural network

A Real-Time Ball Detection Approach Using Convolutional Neural Networks

xYOLO: A Model For Real-Time Object Detection In Humanoid Soccer On Low-End Hardware

Contact Info

Product

Resources

About