BRTS Final Presentation · BRTS Final Presentation 11/29/18. 2 Johnson | Cornell SC Johnson College...
Transcript of BRTS Final Presentation · BRTS Final Presentation 11/29/18. 2 Johnson | Cornell SC Johnson College...
Neuromorphic Vision
BRTS Final Presentation
11/29/18
2 Johnson | Cornell SC Johnson College of Business
Agenda
• Recap midterm feedback
• Getting to target segment
• Culling process
• Developing/testing hypotheses
• Updated business model canvas
• Preliminary market sizing
• Next steps
3 Johnson | Cornell SC Johnson College of Business
Review of Neuromorphic Vision
“Event based” camera only captures movement
Unique to our neuromorphic algorithm?
• Simultaneous independent motion detection
and depth estimation
• Lighter computational weight allows real-time
processing
4 Johnson | Cornell SC Johnson College of Business
Midterm Course Correction
Evaluate potential targets based on assumptions
and preliminary findings along the following criteria:
• Customer Urgency
• Market Size
• Length of Sales Cycle
• Navigating Sales Hurdles
• Technical Difficulty
5 Johnson | Cornell SC Johnson College of Business
Goal: Identify Target Segment
Customer
Urgency
0.3
Market
Size
0.2
Length of
Cycle
0.2
Barriers
To Entry
0.15
Technological
Difficulty
0.15
TOTAL
Wind Farms 2 2 5 4 3 3.05
Ammunition Tracking 3 4 3 1 2 2.75
Incoming Hostile Fire 4 5 3 1 2 3.25
Sports Broadcasting 1 2 4 4 3 2.55
Sports Performance 3 1 4 4 3 2.95
Agricultural Technology 3 3 3 4 4 3.30
Warehouse Automation 2 4 4 4 3 3.25
AV Perception 5 3 3 2 1 3.15
Traffic Planning 1 4 2 2 5 2.55
Camera Technology 2 3 4 4 4 3.20
6 Johnson | Cornell SC Johnson College of Business
Customer Discovery Interviews
• 39 interviews total
• 20 since mid-term presentations
• Ag Tech
• Warehouse Automation
• Camera Tech
• Hostile Fire Detection
• Hypothesis Testing
• Value Seeking
7 Johnson | Cornell SC Johnson College of Business
Application 1 – Ag Tech
• Hypotheses
• Customer Segment: Tree fruit farmers, field crop farmers, high value crop food processors (eg
Heinz and its partners), pesticide companies, disease management analysts
• Value Proposition: 1) Neuromorphic cameras could more quickly, more accurately, and more
cost effectively identify pests, 2) Neuromorphic cameras could more accurately and more cost
effectively automate repetitive high value crop sorting or gathering
• Findings:
• Technologically Feasible– the problems that exist in AgTech are not technically challenging,
though more testing must be done in order to determine feasibility for camera to solve industry
problems
• Many Applications – bird tracking and control, pest identification, field crop AV driving and AV
nut gathering, high value crop sorting
• Identifying the hair on fire issues – bird tracking is seen as a nuisance but current solutions
are seen as sufficient, pest management is expensive and doing this more cheaply or
accurately is valuable though farmer budgets are limited, AV solutions depend on tech
feasibility, high value crop sorting is a hair on fire issue and the biggest potential ag tech market
8 Johnson | Cornell SC Johnson College of Business
Application 2 – Warehouse Automation
• Hypotheses
• Customer segment – e-commerce warehouses, logistics companies
• Value Proposition – for highly automated warehouses, neuromorphic cameras could
reduce lag time and free up computing power for more advanced features that increase
efficiency and decrease error (i.e. AR glasses for pickers, guidance for robotic processes,
2d barcode scanners, inventory movement tracking), creating labor savings
• Findings:
• Not underserved – most significant challenges are largely solved or are being addressed.
Most eager interviewees are product managers at ecommerce retailers, not materials
handling companies. Level of automation varies widely, but is increasing.
• Willing to pay, at scale – In the last few years, Amazon overhauled their picking and
putting system to save on human labor; Amazon has 125k full-time hourly associates
• Opportunities for savings –Even after Amazon’s picking overhaul, users still must hold a
barcode up to a scanner. Neuromorphic vision could shave off time. For sortation systems,
high-speed barcode scanners are expensive (~$100k each), offering additional
opportunities for savings.
9 Johnson | Cornell SC Johnson College of Business
Application 3 – Camera Tech
• Hypotheses
• Customer Segment – find a foothold with independent creatives, action sports gurus, then try to
prove the value to cinema production companies
• Value Proposition – real-time motion detection and depth estimates may create a
tracking/focus system capable of overcoming lack of faith in autofocusing for video
• Findings:
• Willingness to Pay – current cost of neuromorphic sensor is within reason
• Lack of Urgency – inherent distrust of automation in creative and broadcast
• Demands aligned with Tech Strengths – creators value two attributes above all:
• Performance in low light
• Responsiveness
• Few Intermediaries – end user could be primary target audience, direct adopters
• High Compatibility – example rig includes mix-match of 3 or more brands
10 Johnson | Cornell SC Johnson College of Business
Application 4 – Hostile Fire Detection
• Hypotheses
• Customer Segment – Special Forces under Joint Special Operations Command (JSOC)
• Value Proposition – Reduce false positives from existing systems by 95%
• Findings:
• SWAT as early adopters, not JSOC – JSOC required a white paper for discussions. SWAT
frequently uses and tests new weapon technology – adoption timeline would be MUCH faster
than JSOC
• Saving lives – Increase officer safety by determining shooter location, increase civilian safety by
preventing misidentification of the shooter, provide courtroom evidence
• Testbed for ongoing product development – Niagara falls SWAT was willing to provide
shooting range for testing, were very interested in the technology
• Starting with SWAT before approaching Department of Defense is best approach
11 Johnson | Cornell SC Johnson College of Business
• Increase officer
safety by
determining shooter
location
• Increase civilian
safety by preventing
misidentification of
the shooter
• Provide courtroom
evidence
• iniVation
(neuromorphic
camera company)
• Cornell University
• Get: trade show
booths (NYTOA and
NTOA), police
magazines, trials &
demonstrations for
equipment reviewers
• Keep: training
programs
• Software
development
• Hardware integration
• End User Training
• Cameras
• IP
• Access to shooting
range for testing
• Compatibility with
existing equipment
Primary
• Domestic SWAT:
end user – officer,
decision maker –
commander,
purchaser –
purchasing officer
Secondary
• International SWAT
(see above)
• Department of
Defense: end user
–front line military,
decision maker &
purchaser – DOD
CEO or Chiefs of
Staff
• Direct sales to
SWAT
• Direct sales to
Dept. of Defense
• Development & R&D costs (compatibility, upkeep, etc)
• Sales costs
• Camera (if vertical integration to reduce dependency)
• Potential licensing costs if move towards military
• Unit sales
• Maintenance
• Data management or application to process and interpret
information
Hostile Fire Detection
BRTS NMV Team 11/29 2.1Taylor Clawson
12 Johnson | Cornell SC Johnson College of Business
Estimated Opportunity – Near Term (24-48 months)
Total SWAT units = 10,000*
Value Range = $10,000-$30,000
Estimated penetration = 5% in 4 years
Size of Initial Opportunity = $10 million
*Based on survey of International
Association of Chiefs of Police 2013
national SWAT study.
– 17,000 total police units
– 60% surveyed also have SWAT teams
13 Johnson | Cornell SC Johnson College of Business
Total Potential Market
• Best-guess US military market-sizing: $650M
– UAE contract for acoustic detection technology: $52.27M (2015,
DefenseNews)
– UAE total military spend: $23,400M (2017)
– US total military spend: $610,000M (2017)
– Implied US market size: $1,300M
– 50% adjustment for US military buyer power: $650M
14 Johnson | Cornell SC Johnson College of Business
Next Steps
• Test Speeding Bullet Recording/Tracking
• Distribute results in white paper to targeted early adopters
• Secure letters of intent from potential customers
• Additional Customer Discovery
• Develop a pricing model
• Understand the purchase roadmap
• Create a Financial Model
• Threshold for operational profitability
• Estimate acquisition costs
Questions?