Efficient aspect object models using pre-trained convolutional neural networks

Wilkinson, Eric; Takahashi, Takeshi

doi:10.1109/humanoids.2015.7363556

Cited by 3 publications

(3 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This was later extended to a Bayesian model able to provide localization uncertainty , establishing a critical step forward for mobile robots, especially connecting close ties to algorithms that operate under uncertainty. In work by Wilkinson and Takahashi (2015), a pretrained convolutional network was used to predict object descriptions and aspect definitions pertaining to sensory geometries in relation to objects. Unfortunately, class and object descriptors were selected as arbitrary pretrained AlexNet layers and the overall framework relied on a number of thresholds that are difficult to define.…”

Section: Detection Estimation and Trackingmentioning

confidence: 99%

“…The technique localized features corresponding to high activations given point clouds of simple household objects through targeted backpropagation. Using this, they presented a hierarchical controller composing of finger and palm pre-posture positions on the R2 robot, however, alike work by Wilkinson and Takahashi (2015), the specific layer to obtain information from is still human defined.…”

Section: From Perception To Motor Controlmentioning

confidence: 99%

See 1 more Smart Citation

Towards Lifelong Self-Supervision: A Deep Learning Direction for Robotics

Wong¹

2016

Preprint

View full text Add to dashboard Cite

Despite outstanding success in vision amongst other domains, many of the recent deep learning approaches have evident drawbacks for robots. This manuscript surveys recent work in the literature that pertain to applying deep learning systems to the robotics domain, either as means of estimation or as a tool to resolve motor commands directly from raw percepts. These recent advances are only a piece to the puzzle. We suggest that deep learning as a tool alone is insufficient in building a unified framework to acquire general intelligence. For this reason, we complement our survey with insights from cognitive development and refer to ideas from classical control theory, producing an integrated direction for a lifelong learning architecture.

show abstract

Section: Detection Estimation and Trackingmentioning

confidence: 99%

Section: From Perception To Motor Controlmentioning

confidence: 99%

Towards Lifelong Self-Supervision: A Deep Learning Direction for Robotics

Wong¹

2016

Preprint

View full text Add to dashboard Cite

show abstract

“…Operating in embedding space has become an attractive way to deal with highdimensional input, especially when similarity functions are difficult to define on the input space. Robotic sensors such as cameras and lasers have these traits, motivating an array of methods for visual retrieval [2], [24], and recognition [1], [29], [10]. Embedded representations for depth sensors in particular have been used to compute motor commands [23], estimate odometry [21], predict loop closure [17], and even predict the presence of glass [11].…”

Section: Introductionmentioning

confidence: 99%

Laser2Vec: Similarity-based Retrieval for Robotic Perception Data

Nashed¹

2020

2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

As mobile robot capabilities improve and deployment times increase, tools to analyze the growing volume of data are becoming necessary. Current state-of-the-art logging, playback, and exploration systems are insufficient for practitioners seeking to discover systemic points of failure in robotic systems. This paper presents a suite of algorithms for similaritybased queries of robotic perception data and implements a system for storing 2D LiDAR data from many deployments cheaply and evaluating top-k queries for complete or partial scans efficiently. We generate compressed representations of laser scans via a convolutional variational autoencoder and store them in a database, where a light-weight dense network for distance function approximation is run at query time. Our query evaluator leverages the local continuity of the embedding space to generate evaluation orders that, in expectation, dominate full linear scans of the database. The accuracy, robustness, scalability, and efficiency of our system is tested on real-world data gathered from dozens of deployments and synthetic data generated by corrupting real data. We find our system accurately and efficiently identifies similar scans across a number of episodes where the robot encountered the same location, or similar indoor structures or objects.

show abstract

Associating grasp configurations with hierarchical features in convolutional neural networks

Learned-Miller

Grupen

2017

2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

View full text Add to dashboard Cite

Efficient aspect object models using pre-trained convolutional neural networks

Cited by 3 publications

References 15 publications

Towards Lifelong Self-Supervision: A Deep Learning Direction for Robotics

Towards Lifelong Self-Supervision: A Deep Learning Direction for Robotics

Laser2Vec: Similarity-based Retrieval for Robotic Perception Data

Associating grasp configurations with hierarchical features in convolutional neural networks

Contact Info

Product

Resources

About