Active vision in robotic systems: A survey of recent developments

Chen, Shengyong; Li, Youfu; Kwok, Ngaiming

doi:10.1177/0278364911410755

Cited by 357 publications

(209 citation statements)

References 249 publications

Supporting

Mentioning

204

Contrasting

Unclassified

Order By: Relevance

“…While active vision has been researched intensively for over 20 years [3], active touch has been neglected in comparison. However, it is becoming apparent that touch is key to solving some key problems in robotics, e.g.…”

Section: Related Workmentioning

confidence: 99%

Active Bayesian Perception for Simultaneous Object Localization and Identification

Lepora¹,

Martinez-Hernandez²,

Prescott³

2013

Robotics: Science and Systems IX

View full text Add to dashboard Cite

Abstract-In this paper, we propose that active Bayesian perception has a general role for Simultaneous Object Localization and IDentification (SOLID), or deciding where and what. We test this claim using a biomimetic fingertip to perceive object identity via surface shape at uncertain contact locations. Our method for active Bayesian perception combines decision making by threshold crossing of the posterior belief with a sensorimotor loop that actively controls sensor location based on those beliefs. Our findings include: (i) active perception with a fixation control strategy gives an order-of-magnitude improvement in acuity over passive perception without sensorimotor feedback; (ii) perceptual acuity improves as the active control requires less belief to make a relocation decision; and (iii) relocation noise further improves acuity. The best method has aspects that resemble animal perception, supporting wide applicability of these findings.

show abstract

Section: Related Workmentioning

confidence: 99%

Active Bayesian Perception for Simultaneous Object Localization and Identification

Lepora¹,

Martinez-Hernandez²,

Prescott³

2013

Robotics: Science and Systems IX

View full text Add to dashboard Cite

show abstract

“…The most frequently referenced surveys of the field include an overview of early approaches by Scott et al (2003) and an overview of more recent work by Chen et al (2011). We will follow the categorization introduced by Scott et al (2003) in distinguishing between model-based and non-model-based reconstruction methods.…”

Section: Related Workmentioning

confidence: 99%

“…This problem has been well studied in the robotics and computer vision literature Aloimonos et al (1988), Bajcsy (1988), Blake and Yuille (1988), Chen et al (2011), Scott et al (2003), but often an a priori model of the object is assumed, the implementation is robot-dependent, or the sensor pose options are constrained. Based on the current state of the art, it is not clear that there is an optimal way to quantify the volumetric information for the object reconstruction task, with respect to choosing views based on maximizing information gain.…”

Section: Introductionmentioning

confidence: 99%

A comparison of volumetric information gain metrics for active 3D object reconstruction

et al. 2017

View full text Add to dashboard Cite

In this paper, we investigate the following question: when performing next best view selection for volumetric 3D reconstruction of an object by a mobile robot equipped with a dense (camera-based) depth sensor, what formulation of information gain is best? To address this question, we propose several new ways to quantify the volumetric information (VI) contained in the voxels of a probabilistic volumetric map, and compare them to the state of the art with extensive simulated experiments. Our proposed formulations incorporate factors such as visibility likelihood and the likelihood of seeing new parts of the object. The results of our experiments allow us to draw some clear conclusions about the VI formulations that are most effective in different mobile-robot reconstruction scenarios. To the best of our knowledge, this is the first comparative survey of VI formulation performance for active 3D object reconstruction. Additionally, our modular software framework is adaptable to other robotic platforms and general reconstruction problems, and we release it open source for autonomous reconstruction tasks. Abstract In this paper, we investigate the following question: when performing next best view selection for volumetric 3D reconstruction of an object by a mobile robot equipped with a dense (camera-based) depth sensor, what formulation of information gain is best? To address this question, we propose several new ways to quantify the volumetric information (VI) contained in the voxels of a probabilistic volumetric map, and compare them to the state of the art with extensive simulated experiments. Our proposed formulations incorporate factors such as visibility likelihood and the likelihood of seeing new parts of the object. The results of our experiments allow us to draw some clear conclusions about the VI formulations that are most effective in different mobile-robot reconstruction scenarios. To the best of our knowledge, this is the first comparative survey of VI formulation performance for active 3D object reconstruction. Additionally, our modular software framework is adaptable to other robotic platforms and general reconstruction problems, and we release it open source for autonomous reconstruction tasks.

show abstract

“…The problem of computing the optimal views to reconstruct an object or a scene has been studied for more than two decades and is known in the computer vision literature as active vision, View Path Planning (VPP), or Next-Best-View (NBV) [1,2,4,7,21]. Often, the sensor motion is restricted to a sphere and it is assumed that the object of interest is at all times located completely in the sensor frustum.…”

Section: A Related Workmentioning

confidence: 99%

Appearance-based Active, Monocular, Dense Reconstruction for Micro Aerial Vehicles

Förster

Pizzoli

Scaramuzza

2014

Robotics: Science and Systems X

View full text Add to dashboard Cite

Abstract-In this paper, we investigate the following problem: given the image of a scene, what is the trajectory that a robotmounted camera should follow to allow optimal dense depth estimation? The solution we propose is based on maximizing the information gain over a set of candidate trajectories. In order to estimate the information that we expect from a camera pose, we introduce a novel formulation of the measurement uncertainty that accounts for the scene appearance (i.e., texture in the reference view), the scene depth and the vehicle pose. We successfully demonstrate our approach in the case of realtime, monocular reconstruction from a micro aerial vehicle and validate the effectiveness of our solution in both synthetic and real experiments. To the best of our knowledge, this is the first work on active, monocular dense reconstruction, which chooses motion trajectories that minimize perceptual ambiguities inferred by the texture in the scene.

show abstract

Active vision in robotic systems: A survey of recent developments

Cited by 357 publications

References 249 publications

Active Bayesian Perception for Simultaneous Object Localization and Identification

Active Bayesian Perception for Simultaneous Object Localization and Identification

A comparison of volumetric information gain metrics for active 3D object reconstruction

Appearance-based Active, Monocular, Dense Reconstruction for Micro Aerial Vehicles

Contact Info

Product

Resources

About