Obviousness — US Patent 7336805

US patent 7336805, "Docking assistant," describes a method and device for assisting vehicle guidance using image data to identify potential destinations based on their geometric form, without requiring specific markings. The core of the invention lies in a multi-stage exclusionary process for identifying these destinations and then calculating an optimized travel path. An analysis under 35 U.S.C. § 103 reveals that the independent claims (Claim 1 and Claim 13) would likely be considered obvious in light of combinations of prior art references.

The priority date for US7336805 is June 16, 2004. All prior art references discussed below were published before this date.

Obviousness Analysis for Claim 1 (Method)

Claim 1 outlines a method comprising acquiring image data, extracting positional parameters of potential destinations, and calculating an optimized travel path trajectory. The extraction process is detailed, including edge detection/segmentation, storing interrelationships in a mathematical tree structure, analyzing edge segments for geometric objects, analyzing plausibility with a matching algorithm, performing an additional acceptance analysis based on imaging properties, and storing accepted objects.

A combination of the following prior art references would render Claim 1 obvious:

Japanese Patent Application JP 2001-343212 A (Aisin Seiki Co Ltd): This reference describes a camera-based system for guided entry into a parking bay, utilizing visible lane marking signatures. It teaches acquiring image data, identifying visual signatures, measuring their orientation, calculating the vehicle's distance and orientation to the parking bay, and displaying directional arrows to assist the driver. This covers the broad steps of "acquiring image data using an imaging sensor from a surrounding field of the motor vehicle," "extracting from the acquired image data positional parameters of at least one potential destination relative to the motor vehicle," and "calculating at least one trajectory describing an optimized travel path... so as to assist a subsequent vehicle guidance."
German Patent Application DE 10323915.4 (DaimlerChrysler AG): This patent describes a camera-based position detection system for motor vehicles that is "rugged with respect to obscuration and soiling of the visual signatures." It teaches determining the vehicle's position relative to a visual signature (marking a destination) by "matching a template to camera image data acquired from the motor vehicle's surrounding field" and inferring position from "compression and rotation parameters of this template matching." This reference teaches "analyzing the one or more potential geometric objects for plausibility using a matching algorithm."
Robert Sedgewick, "Algorithms in C" (1990, 1998): This non-patent literature provides comprehensive discussions on widely varying tree structures and tree traversal algorithms. The patent itself explicitly references Sedgewick's work when describing the hierarchical tree structure for edge segments. This teaches "storing interrelationships of the plurality of edge segments in a mathematical tree structure" and "analyzing the plurality of edge segments" (using a tree traversal algorithm, as mentioned in the detailed description).
Besl, P. J., McKay, N. D., "A Method for Registration of 3- D Shapes" (1992): This non-patent literature describes the Iterative Closest Point (ICP) algorithm, which the patent specifically mentions as "particularly advantageously suited" for its matching algorithm. It enables objects to be scaled and rotated to minimize quadratic error from an ideal object pattern. This provides a specific example for the "matching algorithm" for plausibility analysis.

Motivation for Combination:

A person having ordinary skill in the art (POSA) seeking to improve the vehicle guidance assistance system of JP '212 (which relies on clearly visible, often soiled or worn, markings) would be motivated to incorporate more robust object detection and recognition techniques. DE '915 offers template matching to make destination identification more resilient to obscuration and soiling, addressing a known drawback of systems like JP '212.

To implement such robust object detection (template matching or 3D shape registration), a POSA would naturally employ standard image processing steps:

Edge detection and segmentation are fundamental preprocessing steps for identifying features suitable for matching, which are well-known in the art (e.g., US4906940A).
Organizing edge segments into a mathematical tree structure (as taught by Sedgewick) is a known method for efficient storage and processing of complex image data, particularly when searching for geometric shapes.
Using a sophisticated 3D matching algorithm like the ICP algorithm (Besl & McKay) would be an obvious choice for accurately determining the 3D position and orientation of identified objects relative to the vehicle, especially if aiming for precise docking.

Regarding the objective of "without the need for affixing specific visual signatures at the destination", a POSA would be motivated to move from systems requiring special markings (like JP '212 and DE '915) to those that can identify inherent geometric forms of destinations (e.g., rectangles, squares for docking stations). Once this objective is adopted, the process of "analyzing the plurality of edge segments for the presence of a geometric object associated with a geometrical form that may at least partially describe a potential destination" becomes an obvious step.

The "additional acceptance analysis" is described in the patent as analyzing "a shape of the image formation of each object in the image data based on knowledge of at least one imaging property of the imaging sensor relative to the surrounding field." This includes recognizing distortions (e.g., a rectangle appearing as a trapezoid from an elevated camera) or using an artificial horizon. A POSA, having integrated an elevated camera (as shown in FIG. 2 of US7336805) and a 3D shape matching algorithm (Besl & McKay), would be well aware of perspective projection effects. It would be an obvious engineering optimization to leverage this knowledge of camera geometry and expected image formation to further filter plausible objects and reduce false positives. For example, if a 3D rectangular object is being sought by an elevated camera, its 2D projection should be a trapezoid. Rejecting candidate objects that do not exhibit the expected trapezoidal distortion for their estimated pose is a logical and obvious refinement to increase the reliability of destination identification. Similarly, using an "artificial horizon" derived from known vehicle features (e.g., loading platform end region 36, as described in US7336805) to filter out objects whose relative position is atypical for a destination (like a cargo door) is a common-sense application of spatial reasoning in computer vision for navigation.

Therefore, the combined teachings provide all elements of Claim 1, with clear motivations for a POSA to combine them to achieve a more robust and versatile vehicle docking assistance system.

Obviousness Analysis for Claim 13 (Device)

Claim 13 describes a device for assisting motor vehicle guidance, comprising an imaging sensor, an image-processing unit (with edge detector/segmenter, locating unit, memory, comparator unit, acceptance analysis unit), a processing unit, and a vehicle guidance system, along with a data memory.

Since Claim 1 (method) is rendered obvious by the combination of prior art, the device claims that perform this method would also be obvious. The construction of a device with functional units corresponding to the steps of an obvious method is a matter of routine engineering for a POSA.

Imaging sensor, processing unit, and vehicle guidance system: These are basic components for any vehicle assistance system, taught by JP '212.
Image-processing unit including an edge detector and segmenter: Edge detection and segmentation hardware/software are standard in image processing systems (e.g., US4906940A).
Locating unit for geometric shapes: Such a unit would be implemented to perform the analysis of edge segments for geometric objects, as taught by the combination for Claim 1.
Memory unit for storing object patterns and a comparator unit for plausibility using a matching algorithm: These components directly implement the template matching (DE '915) or 3D shape registration (Besl & McKay) and pattern storage functions.
Data memory for storing the most proximate destination: This is a standard storage component in any navigation or assistance system.
Acceptance analysis unit: This unit would be implemented to perform the advanced filtering based on imaging properties and expected distortions, as discussed for Claim 1, utilizing existing computer vision hardware/software capabilities.

Motivation for Combination:

The motivation for combining the methods (as detailed above for Claim 1) directly translates into the motivation for configuring a device with the corresponding functional units. A POSA designing a robust vehicle docking assistant would integrate existing hardware components (imaging sensors, processors, memory) and program them (or configure dedicated hardware units) to perform the obvious image processing, object recognition, and guidance tasks. The specific units (edge detector/segmenter, locating unit, comparator unit, acceptance analysis unit) are functional blocks that a POSA would implement using known techniques to achieve the desired improvements in robust, marker-less destination detection.

Therefore, the device claimed in Claim 13, comprising these functionally described units, would be an obvious implementation of the obvious method.