Predicting the grasping point accurately and quickly is crucial for successful robotic manipulation. However, to commercially deploy a robot, such as a dishwasher robot in a commercial kitchen, we also need to consider the constraints of limited usable resources. We present a deep learning method to predict the grasp position when using a single suction gripper for picking up objects. The proposed method is based on a shallow network to enable lower training costs and efficient inference on limited resources. Costs are further reduced by collecting data in a custom-built synthetic environment. For evaluating the proposed method, we developed a system that models a commercial kitchen for a dishwasher robot to manipulate symmetric objects. We tested our method against a model-fitting method and an algorithm-based method in our developed commercial kitchen environment and found that a shallow network trained with only the synthetic data achieves high accuracy. We also demonstrate the practicality of using a shallow network in sequence with an object detector for ease of training, prediction speed, low computation cost, and easier debugging.
Suraj Prakash PATTAR
Connected Robotics Inc.
Tsubasa HIRAKAWA
Chubu University
Takayoshi YAMASHITA
Chubu University
Tetsuya SAWANOBORI
Connected Robotics Inc.
Hironobu FUJIYOSHI
Chubu University
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Suraj Prakash PATTAR, Tsubasa HIRAKAWA, Takayoshi YAMASHITA, Tetsuya SAWANOBORI, Hironobu FUJIYOSHI, "Single Suction Grasp Detection for Symmetric Objects Using Shallow Networks Trained with Synthetic Data" in IEICE TRANSACTIONS on Information,
vol. E105-D, no. 9, pp. 1600-1609, September 2022, doi: 10.1587/transinf.2022EDK0001.
Abstract: Predicting the grasping point accurately and quickly is crucial for successful robotic manipulation. However, to commercially deploy a robot, such as a dishwasher robot in a commercial kitchen, we also need to consider the constraints of limited usable resources. We present a deep learning method to predict the grasp position when using a single suction gripper for picking up objects. The proposed method is based on a shallow network to enable lower training costs and efficient inference on limited resources. Costs are further reduced by collecting data in a custom-built synthetic environment. For evaluating the proposed method, we developed a system that models a commercial kitchen for a dishwasher robot to manipulate symmetric objects. We tested our method against a model-fitting method and an algorithm-based method in our developed commercial kitchen environment and found that a shallow network trained with only the synthetic data achieves high accuracy. We also demonstrate the practicality of using a shallow network in sequence with an object detector for ease of training, prediction speed, low computation cost, and easier debugging.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2022EDK0001/_p
Copy
@ARTICLE{e105-d_9_1600,
author={Suraj Prakash PATTAR, Tsubasa HIRAKAWA, Takayoshi YAMASHITA, Tetsuya SAWANOBORI, Hironobu FUJIYOSHI, },
journal={IEICE TRANSACTIONS on Information},
title={Single Suction Grasp Detection for Symmetric Objects Using Shallow Networks Trained with Synthetic Data},
year={2022},
volume={E105-D},
number={9},
pages={1600-1609},
abstract={Predicting the grasping point accurately and quickly is crucial for successful robotic manipulation. However, to commercially deploy a robot, such as a dishwasher robot in a commercial kitchen, we also need to consider the constraints of limited usable resources. We present a deep learning method to predict the grasp position when using a single suction gripper for picking up objects. The proposed method is based on a shallow network to enable lower training costs and efficient inference on limited resources. Costs are further reduced by collecting data in a custom-built synthetic environment. For evaluating the proposed method, we developed a system that models a commercial kitchen for a dishwasher robot to manipulate symmetric objects. We tested our method against a model-fitting method and an algorithm-based method in our developed commercial kitchen environment and found that a shallow network trained with only the synthetic data achieves high accuracy. We also demonstrate the practicality of using a shallow network in sequence with an object detector for ease of training, prediction speed, low computation cost, and easier debugging.},
keywords={},
doi={10.1587/transinf.2022EDK0001},
ISSN={1745-1361},
month={September},}
Copy
TY - JOUR
TI - Single Suction Grasp Detection for Symmetric Objects Using Shallow Networks Trained with Synthetic Data
T2 - IEICE TRANSACTIONS on Information
SP - 1600
EP - 1609
AU - Suraj Prakash PATTAR
AU - Tsubasa HIRAKAWA
AU - Takayoshi YAMASHITA
AU - Tetsuya SAWANOBORI
AU - Hironobu FUJIYOSHI
PY - 2022
DO - 10.1587/transinf.2022EDK0001
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E105-D
IS - 9
JA - IEICE TRANSACTIONS on Information
Y1 - September 2022
AB - Predicting the grasping point accurately and quickly is crucial for successful robotic manipulation. However, to commercially deploy a robot, such as a dishwasher robot in a commercial kitchen, we also need to consider the constraints of limited usable resources. We present a deep learning method to predict the grasp position when using a single suction gripper for picking up objects. The proposed method is based on a shallow network to enable lower training costs and efficient inference on limited resources. Costs are further reduced by collecting data in a custom-built synthetic environment. For evaluating the proposed method, we developed a system that models a commercial kitchen for a dishwasher robot to manipulate symmetric objects. We tested our method against a model-fitting method and an algorithm-based method in our developed commercial kitchen environment and found that a shallow network trained with only the synthetic data achieves high accuracy. We also demonstrate the practicality of using a shallow network in sequence with an object detector for ease of training, prediction speed, low computation cost, and easier debugging.
ER -