IP Library › Granted Patent US 12,230,649
Granted Patent B2
US 12,230,649 · App. 17/324,964 · Granted Feb 18, 2025

AI system on chip (SOC) for robotics vision applications

Inventor: James E. Retterath (Excelsior, MN)
Assignee: Vergence Automation, Inc.
H01L27/1446G06F9/30101G06N5/04H01L27/14643
View Patent ↗
Loading inventors, assignments & file history…
Monitor This Case
Get email alerts when status or documents change.
Order Certified Copies
Most orders are placed with the USPTO same day — all within 24 business hours.
Order via The Patent Place →
Pre-filled with this patent's details
Quick Facts
Patent No.
US 12,230,649
App. No.
17/324,964
Granted
Feb 18, 2025
Kind
B2
Abstract

An Artificial Intelligence (AI) multi-frame imaging System on Chip (SoC) incorporates in-pixel embedded analog image processing by performing analog image computation within a multi-frame image pixel. In embodiments, each in-pixel processing element includes a photodetector, photodetector control circuitry with at least three analog sub-frame storage elements, analog circuitry configured to process both neighbor-in-space and neighbor-in-time functions for analog data, and a set of north-east-west-south (NEWS) registers, each register interconnected between a unique pair of neighboring in-pixel processing elements to transfer analog data between the pair of neighboring in-pixel processing elements. In embodiments, the in-pixel embedded analog image processing device takes advantage of high parallelism because each pixel has its own processor, and takes advantage of locality of data because all data is located within a pixel or within a neighboring pixel.

Claims (43)

1. A sub-frame imaging pixel comprising:

a photodetector;

photodetector control circuitry (PDC) comprising (i) at least three analog storage elements configured to store at least three sub-frames, wherein each of the at least three sub-frames includes analog data transferred from the photodetector, and (ii) a PDC instruction bus coupled to each of the at least three sub-frames and configured to control operations of the PDC on said analog data transferred from the photodetector; and

an analog pixel processor (APP) configured to process neighbor-in-space and neighbor-in-time functions on said analog data stored in the at least three sub-frames, the APP comprising:

at least two banks of analog registers configured to: (i) receive the analog data from the PDC, and (ii) perform (a) data transfer operations, and (b) one or more of math operations and logic operations on the analog data stored in the at least two banks of analog registers,

a compare-and-flag functional block for each of the at least two banks of analog registers, wherein write operations to a register amongst the at least two banks of analog registers are executed when a signal is active during a register write cycle,

north, east, west, and south (NEWS) registers configured to perform the data transfer operations between neighboring sub-frame imaging pixels to facilitate the neighbor-in-space functions on the analog data stored in the at least two banks of analog registers, and

an APP instruction bus that is separate and distinct from the PDC instruction bus and is configured to: (i) allow for concurrent processing of the analog data stored in two sub-frames of the at least three sub-frames, and (ii) execute each of the following: (a) the data transfer operations, (b) the math operations, (c) the logic operations, (d) the neighbor-in-space functions, and (e) the neighbor-in-time functions.

2. The sub-frame imaging pixel of claim 1 , further comprising:

an analog-to-digital converter (ADC) configured to convert the analog data maintained in one or more of the at least two banks of analog registers to one or more multi-bit digital data values; and

a D/A converter (DAC) configured to convert the one or more multi-bit digital data values to the analog data.

3. The sub-frame imaging pixel of claim 2 , further comprising: a digital storage (SRAM) operably coupled to the ADC and the DAC and configured to cause DAC-converted values to be enabled onto the APP instruction bus and to write ADC-converted values to the SRAM.

4. The sub-frame imaging pixel of claim 1 , wherein:

the at least three analog storage elements implement integration circuitry, and

the PDC further comprises a fourth analog storage element that implements event circuitry.

5. The sub-frame imaging pixel of claim 1 , wherein the at least two banks of analog registers comprise at least one of: (i) four banks of analog registers or (ii) eight banks of analog registers.

6. A semiconductor substrate for sub-frame imaging, including:

an array of sub-frame imaging pixels, wherein each of the sub-frame imaging pixels comprises:

a photodetector;

photodetector control circuitry (PDC) comprising (i) at least three analog storage elements configured to store at least three sub-frames, wherein each the at least three sub-frames includes analog data transferred from the photodetector, and (ii) a PDC instruction bus coupled to each of the at least three sub-frames and configured to control operations of the PDC on said analog data transferred from the photodetector;

an analog pixel processor (APP) configured to process neighbor-in-space and neighbor-in-time functions on said analog data stored in the at least three sub-frames, the APP comprising:

at least two banks of analog registers configured to: (i) receive the analog data from the PDC, and (ii) perform (a) data transfer operations, and (b) one or more of math operations and logic operations on the analog data stored in the at least two banks of analog registers,

a compare-and-flag functional block for each of the at least two banks of analog registers, wherein write operations to a register amongst the at least two banks of analog registers are executed when a signal is active during a register write cycle,

north, east, west, and south (NEWS) registers configured to perform the data transfer operations between neighboring sub-frame imaging pixels to facilitate the neighbor-in-space functions on the analog data stored in the at least two banks of analog registers, and

an APP instruction bus that is separate and distinct from the PDC instruction bus and is configured to: (i) allow for concurrent processing of the analog data stored in two sub-frames of the at least three sub-frames, and (ii) execute each of the following: (a) the data transfer operations, (b) the math operations, (c) the logic operations, (d) the neighbor-in-space functions, and (e) the neighbor-in-time functions;

a photodetector (PD) config memory configured to store information used for sequencing the PDC of each of the array of sub-frame imaging pixels; and

a PDC sequencer configured to step through the information stored in the PD config memory to sequence the PDC of each of the array of sub-frame imaging pixels.

7. The semiconductor substrate of claim 6 , further comprising top-side vias at terminals of the photodetector.

8. The semiconductor substrate of claim 7 , wherein the top-side vias comprise through-silicon vias (TSVs).

9. The semiconductor substrate of claim 6 , wherein a pitch of the sub-frame imaging pixels ranges from 1.5 μm to 40 μm.

10. The semiconductor substrate of claim 6 , wherein a number of sub-frame imaging pixels on a single device is as low as 1024 in a 32×32 grid pattern and as high as 268,435,456 in a 16,384×16,384 grid pattern.

11. A multi-frame imaging system on chip (SoC) comprising:

an array of sub-frame imaging pixels, wherein each of the sub-frame imaging pixels comprises:

a photodetector;

photodetector control circuitry (PDC) comprising (i) at least three analog storage elements configured to store at least three sub-frames, wherein each of the at least three sub-frames includes analog data transferred from the photodetector, and (ii) a PDC instruction bus coupled to each of the at least three sub-frames and configured to control operations of the PDC on said analog data transferred from the photodetector;

an analog pixel processor (APP) configured to process neighbor-in-space and neighbor-in-time functions on said analog data stored in the at least three sub-frames, the APP comprising:

at least two banks of analog registers configured to: (i) receive the analog data from the PDC, and (ii) perform (a) data transfer operations, and (b) one or more of math operations and logic operations on the analog data stored in the at least two banks of analog registers,

a compare-and-flag functional block for each of the at least two banks of analog registers, wherein write operations to a register amongst the at least two banks of analog registers are executed when a signal is active during a register write cycle,

north, east, west, and south (NEWS) registers configured to perform the data transfer operations between neighboring sub-frame imaging pixels to facilitate the neighbor-in-space functions on the analog data stored in the at least two banks of analog registers, and

an APP instruction bus that is separate and distinct from the PDC instruction bus and is configured to: (i) allow for concurrent processing of the analog data stored in two sub-frames of the at least three sub-frames, and (ii) execute each of the following: (a) the data transfer operations, (b) the math operations, (c) the logic operations, (d) the neighbor-in-space functions, and (e) the neighbor-in-time functions; and

a digital processor configured to perform additional processing operations using output from the array of sub-frame imaging pixels.

12. The SoC of claim 11 , wherein the digital processor comprises at least one of a CPU, a GPU, an APU, an FPGA, or an ASIC.

13. The SoC of claim 11 , wherein the digital processor comprises artificial intelligence (AI) software.

Assignments (1)
ASSIGNMENT OF ASSIGNOR'S INTEREST Recorded Jul 27, 2021
From: RETTERATH, JAMES E.
To: VERGENCE AUTOMATION, INC.
Reel/Frame 056995/0446 →
Continuity (2)
Provisional Application 63027227 · May 19, 2020
Related Publication 20210366952A1 · Nov 25, 2021
References Cited (114)
US 6765566B1 · Tsao · 2004 [cited by applicant]
US 7634061B1 · Tumer et al. · 2009 [cited by applicant]
US 8102426B2 · Yahav et al. · 2012 [cited by applicant]
US 8150902B2 · Bates · 2012 [cited by applicant]
US 8543254B1 · Schut et al. · 2013 [cited by applicant]
US 8629387B2 · Pflibsen et al. · 2014 [cited by applicant]
US 9094628B2 · Williams · 2015 [cited by applicant]
US 9185363B2 · Murillo Amaya et al. · 2015 [cited by applicant]
US 9189689B2 · Chandraker et al. · 2015 [cited by applicant]
US 9280711B2 · Stein · 2016 [cited by applicant]
US 9294754B2 · Billerbeck et al. · 2016 [cited by applicant]
US 9325920B2 · Van Nieuwenhove et al. · 2016 [cited by applicant]
US 9367922B2 · Chandraker et al. · 2016 [cited by applicant]
US 9513768B2 · Zhao et al. · 2016 [cited by applicant]
US 9514373B2 · Jeromin et al. · 2016 [cited by applicant]
US 9524434B2 · Gee et al. · 2016 [cited by applicant]
US 9607377B2 · Lovberg et al. · 2017 [cited by applicant]
US 9651388B1 · Chapman et al. · 2017 [cited by applicant]
US 9671243B2 · Stein · 2017 [cited by applicant]
US 9671328B2 · Retterath et al. · 2017 [cited by applicant]
US 9734414B2 · Samarasekera et al. · 2017 [cited by applicant]
US 9778352B2 · Mizutani · 2017 [cited by applicant]
US 9797734B2 · Mizutani et al. · 2017 [cited by applicant]
US 9811731B2 · Lee et al. · 2017 [cited by applicant]
US 9824586B2 · Sato et al. · 2017 [cited by applicant]
US 9836657B2 · Hilldore et al. · 2017 [cited by applicant]
US 9842254B1 · Brailovskiy et al. · 2017 [cited by applicant]
US 9846040B2 · Hallberg · 2017 [cited by applicant]
US 9866816B2 · Retterath · 2018 [cited by applicant]
US 9870513B2 · Thiel et al. · 2018 [cited by applicant]
US 9870624B1 · Narang et al. · 2018 [cited by applicant]
US 9902401B2 · Stein et al. · 2018 [cited by applicant]
US 9905024B2 · Shin et al. · 2018 [cited by applicant]
US 9928605B2 · Bleiweiss et al. · 2018 [cited by applicant]
US 9934690B2 · Kuroda · 2018 [cited by applicant]
US 9940539B2 · Han et al. · 2018 [cited by applicant]
US 9943022B1 · Alam · 2018 [cited by applicant]
US 9946260B2 · Shashua et al. · 2018 [cited by applicant]
US 9953227B2 · Utagawa et al. · 2018 [cited by applicant]
US 9959595B2 · Livyatan et al. · 2018 [cited by applicant]
US 9971953B2 · Li et al. · 2018 [cited by applicant]
US 9981659B2 · Urano et al. · 2018 [cited by applicant]
US 9984468B2 · Kasahara · 2018 [cited by applicant]
US 9992468B2 · Osanai et al. · 2018 [cited by applicant]
US 9996941B2 · Roumeliotis et al. · 2018 [cited by applicant]
US 10012504B2 · Roumeliotis et al. · 2018 [cited by applicant]
US 10012517B2 · Protter et al. · 2018 [cited by applicant]
US 10019014B2 · Prasad et al. · 2018 [cited by applicant]
US 10019635B2 · Kido et al. · 2018 [cited by applicant]
US 10025984B2 · Rajkumar et al. · 2018 [cited by applicant]
US 10037712B2 · Dayal · 2018 [cited by applicant]
US 10046770B1 · Sabri · 2018 [cited by applicant]
US 10049307B2 · Pankanti et al. · 2018 [cited by applicant]
US 10054517B2 · Liu et al. · 2018 [cited by applicant]
US 10055854B2 · Wan et al. · 2018 [cited by applicant]
US 10062010B2 · Kutliroff · 2018 [cited by applicant]
US 10071306B2 · Vandonkelaar · 2018 [cited by applicant]
US 10073531B2 · Hesch et al. · 2018 [cited by applicant]
US 10215856B1 · Xu · 2019 [cited by applicant]
US 10218913B2 · Somasundaram et al. · 2019 [cited by applicant]
US 10302768B2 · Godbaz et al. · 2019 [cited by applicant]
US 10382742B2 · Retterath · 2019 [cited by applicant]
US 10397552B2 · Van Nieuwenhove et al. · 2019 [cited by applicant]
US 20110007160A1 · Okumura · 2011 [cited by applicant]
US 20110285866A1 · Bhrugumalla · 2011 [cited by examiner]
US 20130215290A1 · Solhusvik · 2013 [cited by examiner]
US 20140126769A1 · Reitmayr et al. · 2014 [cited by applicant]
US 20140218480A1 · Knighton et al. · 2014 [cited by applicant]
US 20140347448A1 · Hegemann et al. · 2014 [cited by applicant]
US 20160189372A1 · Lovberg et al. · 2016 [cited by applicant]
US 20160255289A1 · Johnson et al. · 2016 [cited by applicant]
US 20170230638A1 · Rhoads et al. · 2017 [cited by applicant]
US 20170236037A1 · Wajs et al. · 2017 [cited by applicant]
US 20170323429A1 · Godbaz et al. · 2017 [cited by applicant]
US 20180031681A1 · Yoon et al. · 2018 [cited by applicant]
US 20180063508A1 · Trail et al. · 2018 [cited by applicant]
US 20180113200A1 · Steinberg et al. · 2018 [cited by applicant]
US 20180176514A1 · Kirmani et al. · 2018 [cited by applicant]
US 20180188059A1 · Wheeler et al. · 2018 [cited by applicant]
US 20180330526A1 · Corcoran · 2018 [cited by applicant]
US 20180358393A1 · Sato · 2018 [cited by examiner]
US 20190033448A1 · Molnar et al. · 2019 [cited by applicant]
US 20190056498A1 · Sonn et al. · 2019 [cited by applicant]
US 20190230297A1 · Knorr et al. · 2019 [cited by applicant]
US 20190286153A1 · Rankawat et al. · 2019 [cited by applicant]
US 20200057146A1 · Steinkogler et al. · 2020 [cited by applicant]
US 20200278194A1 · Kawahito · 2020 [cited by applicant]
DE 102018107801A1 · 2018 [cited by applicant]
KR 1020160135482A · 2016 [cited by applicant]
WO WO2018127789A1 · 2018 [cited by applicant]
WO WO2020198134A1 · 2020 [cited by applicant]
Krestinskaya, O. and Pappachen James, A., “Real-time Analog Pixel-to-pixel Dynamic Frame Differencing with Memristive Sensing Circuits”, arXiv e-prints, arXiv:1808.06780v1 [cs.ET], Aug. 21, 2018, doi: 10.48550/arXiv.180… [cited by examiner]
P. Dudek and P. J. Hicks, “A general-purpose processor-per-pixel analog SIMD vision chip,” in IEEE Transactions on Circuits and Systems I: Regular Papers, vol. 52, No. 1, pp. 13-20, Jan. 2005, doi: 10.1109/TCSI.2004.840… [cited by examiner]
Abuelsamid, “Bosch Launches Silicon Carbide Chips to Make Evs More Efficient,” Forbes, Oct. 13, 2019, (accessed at https://www.forbes.com/sites/samabuelsamid/2019/10/13/bosch-launches-silicon-carbide-chips-to-make-evs-m… [cited by applicant]
Barati et al., “Hot Carrier-Enhanced Interlayer Electron-Hole Pair Multiplication in 2D Semiconductor Heterostructure Photocells,” Nature Nanotechnology, Oct. 9, 2017, 7 pages. [cited by applicant]
Becker et al., “Smartphone Video Guidance Sensor for Small Satellites,” NASA Marshall Space Flight Center, 27th Annual AIAA/USU Conference on Small Satellites, Aug. 2013, 8 pages. [cited by applicant]
Bie et al., A MoTe2-Based Light-Emitting Diode and Photodetecto for Silicon Phototonic Integated Circuits, Nature Nanotechnology, Oct. 23, 2017, 8 pages. [cited by applicant]
Dionne et al., “Silicon-Based Plasmonics for On-Chip Phototonics,” IEEE Journal, Jan.-Feb. 2010, 13 pages. [cited by applicant]
Dudek et al., “A General-Purpose CMOS Vision Chip with a Process-Per-Pixel SIMD Array,” Computer Science, 2001, 4 pages. [cited by applicant]
Dudek, “Adaptive Sensing and Image Processing with a General-Purpose Pixel-Parallel Sensor/Processor Array Integrated Circuit,” University of Manchester, 2006, 6 pages. [cited by applicant]
Evans, “Cascading Individual Analog Counters,” Radiant Technologies, Inc., Oct. 2016, 7 pages. [cited by applicant]
Foix et al., “Exploitation of Time-of-Flight (ToF) Camera,” IRI Technical Report, 2007, 22 pages. [cited by applicant]
Hall et al., “Guide for Pavement Friction,” NCHRP, National Academies of Sciences, Engineering, and Medicine, 2009, 257 pages. [cited by applicant]
Martel et al., “Parallel HDR Tone Mapping and Auto-Focus on a Cellular Processor Array Vision Chip,” 2016 IEEE International Symposium on Circuits and Systems, May 2016, 4 pages. [cited by applicant]
Panina et al., “Compact CMOS Analog Counter for SPAD Pixel Arrays,” IEEE, Apr. 2014, 5 pages. [cited by applicant]
Peizerat et al., “An Analog Counter Architecture for Pixel-Level ADC,” CEA/LETI-MINATEC, 2009, 3 pages. [cited by applicant]
Pinson et al., “Orbital Express Advanced Video Guidance Sensor: Ground Testing, Flight Results and Comparisons,” NASA Marshall Space Flight Center, American Institute of Aeronautics and Astronautics, Aug. 2008, 12 pages. [cited by applicant]
Sun et al., “Single-Chip Microprocessor that Communicates Directly Using Light,” Nature, Dec. 23, 2015, 29 pages. [cited by applicant]
Tang et al., “2D Materials for Silicon Photonics,” Nature Nanotechnology, Oct. 23, 2017, 2 pages. [cited by applicant]
Torrens, “4QD-TEC: Electronics Circuits Reference Archive Analogue Pulse Counter,” [undated], 2 pages. [cited by applicant]
University of Bonn., “Teaching Cars to Drive with Foresight: Self-Learning Process,” Science Daily, Oct. 2019, 4 pages. [cited by applicant]
Vijayaraghavan et al., “Design for MOSIS Educational Program (Research),” EE Department, University of Tennessee, [undated], 8 pages. [cited by applicant]
Wong et al., “Analog Vision—Neural Network Inference Acceleration Using Analog SIMD Computation in the Focal Plane,” Imperial College London, Diplomarbeit, 2018, 112 pages. [cited by applicant]
Wong, the SCAMP-5 Vision Chip is a Focal-Plane Sensor-Processor (FPSP) developed at the University of Manchester (Carey et al., 2013a), 6 pages. [cited by applicant]