Vision and Dead Reckoning-based End-to-End Parking for Autonomous Vehicles

Rathour, Swarn Singh; John, Vijay; Nithilan, M. K.; Mita, Seiichi

doi:10.1109/ivs.2018.8500558

Cited by 8 publications

(6 citation statements)

References 13 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Other state-of-art methods [7] use machine learning for locating and guiding for autonomous parking. Rathour et al [8] proposed a deep learning network which operates on front and rearview camera data and provides steering angle for end-to-end parking. Deep learning based general object detection techniques such as YOLO [9,10] can help in detecting objects in parking slots.…”

Section: Fig 1 Sample Parking Slotsmentioning

confidence: 99%

Camera Based Parking Slot Detection for Autonomous Parking

Suddamalla

Wong

Balaji

et al. 2021

Communications in Computer and Information Science

View full text Add to dashboard Cite

Autonomous parking is one of the primary functionalities of intelligent vehicles. This requires the accurate detection of a free parking slot and its dimensions. Here we propose a method using a novel combination of computer vision techniques on the camera data mounted on sides of a car. Our occupancy check method is based on perspectivity guided motion segmentation, which makes it a generic approach for handling random obstacles without prior knowledge. When tested on our dataset collected in environments such as shadows, wet surface, indoors and outdoors with obstacles including cars, motorbikes, cones, carton boxes and trees, this method achieved a promising performance with F1 score higher than 97%. With the ability to run on low computational devices such as CPU, this method is adaptable to practical solutions for both AD and aftermarket ADAS systems.

show abstract

Section: Fig 1 Sample Parking Slotsmentioning

confidence: 99%

Camera Based Parking Slot Detection for Autonomous Parking

Suddamalla

Wong

Balaji

et al. 2021

Communications in Computer and Information Science

View full text Add to dashboard Cite

show abstract

“…By learning the control laws between the vehicle's states and the corresponding actions, the proposed ANN-based controller could yield a steering angle and the adjusted desired velocity to complete automatic parking in a confined space.Fuzzy-based and knowledge-based approaches have been presented [17][18][19]. These methods are capable of providing solutions within a range of designed rules with an advantage in easy implements and practical usages.Recently, methods based on deep neural networks have been expected to solve the drawbacks of the mentioned step-by-step approaches by maneuvering vehicles without prior offline trajectory planning [20][21][22][23]. By training an artificial neural network (ANN) using a dataset generated by simulation or experiment, the ANN learns hyper-dimensional relationships between the current vehicle states and the appropriate vehicle maneuvering signals.…”

mentioning

confidence: 99%

“…Li et al proposed an end-to-end neural-network-based automatic parking controller [21]. Rathour also proposed an encoder-decoder architecture for automatic parking [22]. Moon et al developed an automatic parking controller with a twin ANN architectures [23].However, neither the planning-based methods nor ANN-based controllers have taken account of longitudinal control delay for a vehicle under automatic parking.…”

mentioning

confidence: 99%

“…In planning-and-tracking methods, the parking controller maneuvers according to a predetermined trajectory as the reference path. Meanwhile, in ANN-based methods, the parking controller is trained using various deep learning techniques [20][21][22][23]. Longitudinal latency, the delay time between the input velocity and the real velocity of the vehicle, is unavoidable in vehicle dynamics control.…”

mentioning

confidence: 99%

“…Recently, methods based on deep neural networks have been expected to solve the drawbacks of the mentioned step-by-step approaches by maneuvering vehicles without prior offline trajectory planning [20][21][22][23]. By training an artificial neural network (ANN) using a dataset generated by simulation or experiment, the ANN learns hyper-dimensional relationships between the current vehicle states and the appropriate vehicle maneuvering signals.…”

mentioning

confidence: 99%

See 2 more Smart Citations

An Inverse Vehicle Model for a Neural-Network-Based Integrated Lateral and Longitudinal Automatic Parking Controller

2019

View full text Add to dashboard Cite

The majority of currently used automatic parking systems exploit the planning-and-tracking approach that involves planning the reference trajectory first and then tracking the desired reference trajectory. However, the response delay of longitudinal velocity prevents the parking controller from tracing the desired trajectory because the vehicle's velocity and other state parameters are not synchronized, while the controller maneuvers the vehicle according to the planned desired velocity and steering profiles. We propose an inverse vehicle model to provide a neural-network-based integrated lateral and longitudinal automatic parking controller. We approximated the relationship of the planned velocity to the vehicle's velocity using a second-order difference equation that involves the response characteristic of the vehicle's longitudinal delay. The adjusted desired velocity to track the origin-planned velocity is calculated using the inverse vehicle model. Furthermore, we proposed an integrated longitudinal and lateral parking controller using an artificial neural network (ANN) model trained on a dataset applying the inverse vehicle model. By learning the control laws between the vehicle's states and the corresponding actions, the proposed ANN-based controller could yield a steering angle and the adjusted desired velocity to complete automatic parking in a confined space.Fuzzy-based and knowledge-based approaches have been presented [17][18][19]. These methods are capable of providing solutions within a range of designed rules with an advantage in easy implements and practical usages.Recently, methods based on deep neural networks have been expected to solve the drawbacks of the mentioned step-by-step approaches by maneuvering vehicles without prior offline trajectory planning [20][21][22][23]. By training an artificial neural network (ANN) using a dataset generated by simulation or experiment, the ANN learns hyper-dimensional relationships between the current vehicle states and the appropriate vehicle maneuvering signals. Instead of calculating the parking trajectory offline, the ANN-based parking controller can yield a direct maneuvering signal of the steering angle and velocity online, while the vehicle is moving into a parking space. Liu et al. presented a method to enumerate all the possible parking trajectories and corresponding steering actions, and then have the parking controller learn the relationship between the given initial-and-final state pairs and the corresponding sequence of steering actions using an ANN [20]. Li et al. proposed an end-to-end neural-network-based automatic parking controller [21]. Rathour also proposed an encoder-decoder architecture for automatic parking [22]. Moon et al. developed an automatic parking controller with a twin ANN architectures [23].However, neither the planning-based methods nor ANN-based controllers have taken account of longitudinal control delay for a vehicle under automatic parking. Figure 1 shows a typical block diagram of a conventional automatic parking contr...

show abstract