Methods and devices for generating a representation of a 3D scene at very high speed

• US 9,516,244 B2
• Filed: 12/21/2011
• Issued: 12/06/2016
• Est. Priority Date: 12/23/2010
• Status: Active Grant

• Current Assignee: Fastree3D SA
• Original Assignee: Fastree3D SA
• Inventors: Borowski, Andr
• Primary Examiner(s): Ratcliffe, Luke
• Application Number:
  US13/995,654
• Publication Number:
  US 20130300838A1
• Time in Patent Office: 1,812 Days
• Field of Search: 356 301- 515
• US Class Current: 1/1
• CPC Class Codes: G01S 17/90 using synthetic aperture techniques
• CPC Class Codes: G01S 7/486 Receivers
• CPC Class Codes: G01S 7/4863 Detector arrays, e.g. charge-transfer gates
• CPC Class Codes: G01S 7/4865 Time delay measurement, e.g. time-of-flight measurement, time of arrival measurement or determining the exact position of a peak peak detection in noise, signal conditioning G01S7/487
• CPC Class Codes: G01S 7/4868 Controlling received signal intensity or exposure of sensor
• CPC Class Codes: H04N 25/00 Circuitry of solid-state image sensors [SSIS]; Control thereof
• CPC Class Codes: H04N 25/773 comprising photon counting circuits, e.g. single photon detection [SPD] or single photon avalanche diodes [SPAD]

Abstract

The present invention relates to a 3D landscape real-time imager. It also relates to methods for operating such an imager. Such an imager comprises: —at least one illuminating part which is designed to scan at least a portion of the landscape at a given range and having an ultra-short laser pulse source emitting at least one wavelength, and an optical rotating block, with a vertical axis of rotation, and controlled such that given packets of pulses are shaped in a pattern of rotating beams sent toward the said at least partial landscape; —at least one receiving part which comprises a set of SPAD detector arrays, each arranged along a vertical direction and rotating at a given speed in synchronism with the optical rotating block of the illuminating part, the detection data of the SPAD detector arrays being combined to acquire 3D imaging data of the said at least partial landscape in a central controller.

Claims

1. A method for producing a representation of a scene by means of a sensor array, comprising:

illuminating the scene by means of a series of light pulses, wherein an arrival time of each of the light pulses is measured as a timestamp;

measuring a distance with the scene;

collecting light reflected or scattered by the scene to at least a single photon avalanche diode (SPAD) sensor array, comprising at least one SPAD detector cell, on the basis of the pulses from said series of light pulses such that a reference time is determined first, and then a time of flight of each event detected at each SPAD detector cell is counted from the determined reference time, the reference time being unique for at least a train of a predetermined series of light pulses; and

generating one of a 2D or 3D representation of the illuminated scene by processing the electrical signals generated by each SPAD detector cell on the basis of the counted time of flight of the events,

such that said representation allows to detect movements in the scene and/or to achieve high resolution 2D or 3D images, wherein a time separation between the light pulses is shorter than the time of flight.

2. The method according to claim 1, wherein the timestamp origin is defined by the first pulse of the series of light pulses.

3. The method according to claim 1, wherein the timestamp is a digital word comprising coarse and fine parts, wherein the coarse part is obtained by counting a clock edges and wherein the timestamps are averaged.

4. The method according to claim 1, wherein the received pulses are registered by at least one SPAD detector cell and/or a plurality of said SPAD detector cells at one time or at different times.

5. The method according to according to claim 1, wherein said generating step comprises a selection of at least one of the following steps of:

using a global timestamp for each detected event on at least a SPAD detector cell on a frame of the pulsed train;

using a variable number of pulses for each measurement pulse train, that number being optimized considering the speed of measurement and the obtained image resolution while keeping the involved optical energy low enough for obtaining eye safety;

filtering the events detected at each SPAD detector cell;

having generated a plurality of packets of a given number of laser pulses, averaging, notably under a square law, the received packets at least upon one SPAD;

locally close to the SPAD detector, compressing of a time data before further processing;

generating a clock and power enabling signal on the basis of a first coarse scanning by an illuminating device to activate at least one SPAD detector cell just-in-time, calculated from 3D data previously acquired, for the arrival of a packet of received laser pulses;

generating a series of laser pulses arranged in tagged packets of pulses and detecting tags on said received light pulses in SPAD detector cells;

detecting and/or filtering of the time and depth values from at least one pixel of the scene from different pulses diffused toward available SPAD detector cells;

acquiring the same approximate pixel address in the scene on the basis of at least two wavelengths reaching successively different wavelength-filtered SPAD detectors within a time which is shorter than a single SPAD'"'"'s recovery time;

acquiring illuminated lines of the scene;

acquiring illuminated lines of the scene on a single line focal point, then diffused onto multiple SPAD detector arrays and then accumulating/averaging the detected time-related data to generate a composite pixel in a purely digital way;

just-in-time powering and clocking of SPAD for power saving.

6. The method according to according to claim 1, wherein it comprises a step of performing a multi-wavelength strategy comprising:

a. emitting at least two separate laser pulses at different wavelengths; and

b. receiving the reflected light pulses at different detectors sensitive to one wavelength.

7. A method for making a SPAD detector array to be used in a method for producing a representation of a scene by means of a sensor array and comprising:

illuminating the scene by means of a series of light pulses, wherein an arrival time of each of the light pulses is measured as a timestamp;

collecting light reflected or scattered by the scene to at least a single photon avalanche diode (SPAD) sensor array comprising at least one SPAD detector cell, on the basis of the pulses from said predetermined series of light pulses such that a reference time is determined first, and then a time of flight of each event detected at each SPAD detector cell is counted from the determined reference time, the reference time being unique for at least a train of a predetermined series of light pulses; and

generating one of a 2D or 3D representation of the illuminated scene by processing the electrical signals generated by each SPAD detector cell on the basis of the counted time of flight of the events,

such that said representation allows to detect movements in the scene and/or to achieve high resolution 2D or 3D images, wherein a time separation between the light pulses is shorter than the time of flight,

said method for making a SPAD detector array comprising the steps of;

making a first layer comprising a substrate; and

then

making a second layer comprising a plurality of single SPAD cells and using a semiconductor body, at least some SPAD cells having also local analog and/or timing signal processing and also digital signal processing.

8. The method according to according to claim 7, wherein the step of making a second SPAD detector layer comprises the steps of:

making a first CMOS layer at least partially as the first layer comprising

depositing a germanium layer as a semiconductor body by growing a single grain onto a single SPAD cell;

said second SPAD detector layer being formed under a μ

-Czochralski method.

9. The method according to according to claim 8, wherein said SPAD cell comprises at least a single photon avalanche diode and the method comprises:

a step of forming a surrounding cathode under the limit of the grain boundary of said Ge layer;

a step of forming a central pad region forming an anode with p+/n+ region at the central hole.

10. The method according to according to claim 9, wherein said central hole has been formed to be a photon detection device by setting its horizontal and/or vertical dimensions to absorb and register a single photon at said wavelength range.

11. The method according to according to claim 8, wherein said SPAD detector cell comprises at least a CMOS transistor formed with:

an gate electrode implemented onto the central hole forming a channel region of said CMOS transistor;

an source electrode formed onto a p+/n+ region at a first side of said channel region; and

an drain electrode formed onto a p+/n+ region at a second side of said channel region;

said CMOS transistor belonging to a local digital and/or analog signal processing device built on said SPAD detector cell.

12. A single photon avalanche diode (SPAD) detector array, comprising:

at least a SPAD detector cell having at least a SPAD diode;

an optical diffuser distributing randomly photons on several stacked SPADs detector cells in a line array detector; and

at least one of the following devices;

a quenching circuit,

a pulse conditioner,

at least one time-to-digital converter,

a local digital signal processor,

a blanking/filtering circuit controlled by said local digital signal processor,

a statistical processing circuit comprising at least one counter of the detected events at the SPAD diode, and a D-latch and a multiplier connected to an inverse calculus and eventually to a local correcting circuit,

a digital time and/or space averaging circuit,

a digital nonlinear averaging circuit, and

a time/space data compression circuit.

13. The SPAD detector array according to claim 12, wherein it further comprises an optical diffuser combined with several SPADs detector cells in a matrix detector.

14. The SPAD detector array according to claim 12, wherein at least some of the SPAD detector cells are arranged as macrocells, having a macrocell digital signal processor devoted to control and process the events of the associated SPAD detector cells of the macrocell.

15. The SPAD detector array according to claim 12, wherein it comprises a local digital signal processor at the SPAD detector cell level, a local digital signal processor at each group of SPAD detector cell level and at least one digital signal processor at the global level, each level of digital signal processors executing some programs devoted to the detected events or digital data generated at the closest part and to limit the data transfer requirements and means for building on-demand a flexible, local macro-cell.

16. The SPAD detector array according to claim 14, wherein it comprises a local digital signal processor at the SPAD detector cell level, a local digital signal processor at each group of SPAD detector cell level and at least one digital signal processor at the global level, each level of digital signal processors executing some programs devoted to the detected events or digital data generated at the closest part and to limit the data transfer requirements and means for building on-demand a flexible, local macro-cell.

17. A device for the representation of a scene, comprising:

at least an illuminating device of said scene providing at least a series of ultrashort laser pulses having a duration in the range of a few picoseconds with time-related positions under control of a controller, the illuminating device comprising at least a mode-locked laser source connected to an erbium-doped optical fiber amplifier; and

at least a single photon avalanche diode detector array arranged to detect the reflected light from a scene from at least said series of ultrashort laser pulses of said illuminator,

wherein a reference time is determined, and then a time of flight of each event detected at the detector array is counted from the determined reference time, the reference time being unique for at least a train of a predetermined series of laser pulses, and a time separation between the laser pulses is shorter than the time of flight.