PM Portfolio Manager Reports

From Portfolio Manager reports

Perception, Tactile Sensing, And Localization

This node covers the hardware and software that lets a robot see objects, estimate position, measure distance, detect contact, and verify task completion. The stack includes industrial machine vision, cameras, lidar, radar, sonar, IMUs, tactile and force sensors, e-skin, edge vision processors, localization software, inspection systems, and safety sensing. The paper prior says robots need real-time multimodal sensing before they can work safely near people and flexible parts. The local knowledge posterior supports public-market exposure first through machine vision and scaled 3D sensing, then through edge perception compute and specialty sensing. No ranked company gets direct humanoid revenue credit without source-routed disclosure.

Current Setup

Sensing conversion gate Public evidence is strongest in machine vision and 3D sensing; tactile remains the gap.

Ten xhigh primary agents and one xhigh merge agent ranked source-backed node exposure first, operating evidence second, and chart timing third. The final core is CGNX, HSAI, TDY, OUST, and AMBA, with CODA labeled as a specialty watch.

Paid exposure CGNX and TDY already monetize machine vision, imaging, and sensing budgets. These are the best filed routes to industrial vision and high-reliability imaging, but their reports still reflect broad automation, logistics, defense, and industrial cycles.
3D sensing HSAI has the best disclosed lidar scale; OUST has the cleaner U.S.-listed lidar route. ADAS and smart-infrastructure evidence can harden production, but it does not equal humanoid revenue. Product margin, receivables, and cash flow decide the next read.
Tactile gap LINK is the tactile watch, while Keyence, Omron, SICK, and Hikrobot stay source-only. Touch and e-skin are important in the papers, but local filed public-company coverage is still thin and subscale.
  1. Paper priorRobots need scene, pose, force, and contact data
  2. QualificationSensor stacks win OEM and factory programs
  3. ShipmentDesign wins become product revenue and units
  4. EconomicsGross margin, receivables, and cash flow hold
  5. Refresh triggerNew filings or chart breaks change the ranking

Basket

The ranked cards below follow the merge result from ten xhigh primary research agents and one xhigh tournament selector. The ranking uses filed node exposure and operating proof before chart timing. Local daily OHLC coverage exists through 2026-07-02 for all listed tickers. The right rail chart reads the report API when it is running; this page does not embed static OHLC data or horizontal chart levels.

CGNX core machine vision

Clearest public route to paid industrial machine-vision exposure for object recognition, inspection, guidance, ID, and factory workflows.

Market cap$11.3B
Next earningsNot confirmed
Latest qtr revenue$268.4M

Role in stack

CGNX sells machine-vision systems, sensors, ID readers, and software used to inspect, identify, guide, and verify work in factories and logistics. It is the cleanest public proxy for the mature vision layer.

Revenue mix

Q1 2026 revenue was $268.4M, up 24%, with 71.1% gross margin, 22.3% operating margin, $42M of free cash flow, $621.9M of cash and investments, and no debt.

Proof burden

CGNX needs Q2 revenue of $280M-$300M, 28%-31% adjusted EBITDA margin, broad order quality, and AI product adoption to support the rerating and reduce one-customer timing risk.

HSAI core lidar scale

Best disclosed lidar unit scale and robotics-lidar shipment evidence, with high production relevance and policy, receivable, and ASP caveats.

Market cap$2.7B
Next earningsNot confirmed
Latest qtr revenueRMB680.6M

Role in stack

HSAI supplies lidar for ADAS and robotics applications. Its ADAS scale hardens production and cost learning, while robotics shipments make it a direct 3D sensing candidate for this node.

Revenue mix

Q1 2026 revenue was RMB680.6M with 471,723 lidar shipments, 39.1% gross margin, positive core lidar operating income, and robotics shipments up 137.8% year over year.

Proof burden

HSAI must hold Q2 revenue and shipment guidance, protect gross margin as ASPs fall, collect receivables, limit SGI losses, and contain Section 1260H procurement risk.

TDY core imaging scale

Scaled imaging, infrared, instrumentation, and defense sensing exposure with strong operating quality and indirect humanoid attribution.

Market cap$30.2B
Next earningsNot confirmed
Latest qtr revenue$1.560B

Role in stack

TDY brings Teledyne FLIR, digital imaging, infrared detectors, surveillance, unmanned-system, instrumentation, and aerospace/defense electronics exposure to the sensing stack.

Revenue mix

Q1 2026 sales were $1.560B, up 7.6%. Digital Imaging grew 7.9%, Aerospace and Defense Electronics grew 14.4%, RPO reached $4.867B, and Q1 free cash flow was $204.3M.

Proof burden

TDY needs Digital Imaging and ADE orders to convert while free cash flow recovers after inventory build. The page treats it as sensing and imaging exposure, not disclosed humanoid revenue.

OUST core U.S. lidar

Clean U.S.-listed lidar and 3D perception route with product revenue, sensor shipments, industrial relevance, and stronger current timing.

Market cap$3.2B
Next earningsNot confirmed
Latest qtr revenue$48.6M

Role in stack

OUST sells digital lidar, 3D sensing, and perception products for industrial, robotics, smart infrastructure, automotive, and defense use cases. Stereolabs adds 3D vision breadth.

Revenue mix

Q1 2026 revenue was $48.6M, product revenue was $48.2M, shipments were above 12,600 sensors, Q2 guidance was $49.5M-$52.5M, and cash/investments were $174.9M with no debt.

Proof burden

OUST needs product gross margin, adjusted EBITDA, operating cash flow, customer breadth, Stereolabs contribution, and share-count discipline to catch up with the strong chart.

AMBA core edge vision compute

Edge-AI perception compute route for cameras, embedded vision, automotive/IoT systems, and robotics design wins.

Market cap$3.4B
Next earningsNot confirmed
Latest qtr revenue$100.4M

Role in stack

AMBA supplies low-power edge-AI vision SoCs and perception compute that can process camera and sensor data at the endpoint. It belongs to this node through sensor fusion and vision inference.

Revenue mix

FY2026 revenue was $390.7M, up 37.2%, with edge AI about 80% of revenue. Q1 FY2027 revenue was $100.4M, and the lane cites robotics design wins and more than 46M AI SoCs installed.

Proof burden

AMBA needs Q2 FY2027 revenue of $105M-$111M, inventory conversion, customer concentration improvement, design-win revenue, and a clearer path through GAAP losses and SBC.

CODA specialty sensing watch

Profitable acoustic and sonar sensing row with strong niche economics, but its defense/subsea evidence is a weak humanoid perception fit.

Market cap$109M
Next earningsNot confirmed
Latest qtr revenue$6.9M

Role in stack

CODA supplies sonar, acoustic imaging, DAVD, and specialty defense/subsea sensing. It is a source-backed physical sensing company, but its route is specialty localization and acoustic perception.

Revenue mix

Q2 FY2026 revenue was $6.9M with 66.3% gross margin, $1.8M operating income, $30.6M of cash, and no revolver draw. Marine Technology was 41.1% of revenue and fell 26.8% year over year.

Proof burden

CODA needs Marine recovery, DAVD/NANO orders, backlog visibility, repeat procurement, and a clearer localization use case before it can move from specialty watch to core perception exposure.

AEVA watch 4D lidar

Best non-core lidar/localization watch because FMCW 4D lidar is directly relevant and local timing is above short and long trend.

Market cap$1.6B
Next earningsNot confirmed
Latest qtr revenue$6.3M

Role in stack

AEVA is a 4D FMCW lidar candidate for range, velocity, and localization. The local lane cites customer validation and program references, but product revenue is still early.

Revenue mix

Q1 2026 revenue was $6.3M, with product revenue of $2.4M and professional services/NRE of $3.8M. Cash and securities were $99.5M, and FY2026 revenue target was $30M-$36M.

Proof burden

AEVA needs product revenue to replace NRE-heavy growth, program milestones to become shipments, and burn/financing terms to improve before it deserves core treatment.

LINK watch tactile proxy

Closest public tactile and force-sensing proxy, but scale, liquidity, customer concentration, and cash runway keep it out of core.

Market cap$67M
Next earningsNot confirmed
Latest qtr revenue$3.1M

Role in stack

LINK sells force/touch HMI, force-sensing resistors, printed electronics, membrane keypads, and environmental sensors. It is the best filed local route to tactile sensing.

Revenue mix

Q1 2026 revenue was $3.1M, gross margin was 43.5%, operating loss was $0.45M, operating cash flow was negative $0.54M, and cash was $2.1M.

Proof burden

LINK needs repeat revenue, positive operating cash flow, customer diversification, cash stability, and explicit robotics, hand, or end-effector force/touch evidence.

INVZ watch lidar

Automotive lidar and perception software option with customer references, but weaker current revenue, margin, liquidity, and timing.

Market cap$149M
Next earningsNot confirmed
Latest qtr revenue$7.1M

Role in stack

INVZ provides lidar and perception software with automotive and physical-AI relevance. It stays watch because customer references have not yet produced strong current economics.

Revenue mix

Q1 2026 revenue was $7.1M, down from $17.4M, gross margin was negative 22.2%, liquidity was $60.1M, and operating cash flow was negative $14.5M.

Proof burden

INVZ needs NRE catch-up, FY2026 revenue guidance delivery, positive gross margin, lower cash burn, and slower dilution before it can move above watch status.

ARBE watch radar

Radar can support localization, safety, and perception redundancy, but current revenue and gross profit are still too small.

Market cap$96M
Next earningsNot confirmed
Latest qtr revenue$0.46M

Role in stack

ARBE is a radar/localization option for safety, redundancy, and all-weather sensing. It is relevant to perception architecture but remains pre-scale.

Revenue mix

FY2025 revenue was $1.0M, Q1 2026 revenue was $0.46M, backlog was $1.0M, and Q1 adjusted EBITDA loss was $9.9M with $53.6M of cash and deposits.

Proof burden

ARBE needs backlog conversion, positive gross profit, revenue toward the $4M-$6M FY2026 guide, burn reduction, and Nasdaq compliance.

ODYS watch visual AI

Visual AI and predictive-maintenance route with defense/aerospace validation, but it is adjacent to robot perception and very early in revenue conversion.

Market cap$67M
Next earningsNot confirmed
Latest qtr revenue$0.08M

Role in stack

ODYS brings visual monitoring, inspection, and predictive-maintenance software context. It can inform industrial sensing, but the inspected evidence is not onboard humanoid sensing.

Revenue mix

Q1 2026 revenue was $0.082M, FY2025 revenue was $3.0M, backlog was $14M, cash was $21.8M, and operating cash flow was negative $4.3M.

Proof burden

ODYS needs backlog-to-revenue conversion, gross margin proof, customer diversification beyond concentrated programs, and ATM discipline.

MSAI watch monitoring software

Multi-sensor monitoring and software exposure that fits factory reliability, but current evidence is small, concentrated, and peripheral to humanoid sensing.

Market cap$10M
Next earningsNot confirmed
Latest qtr revenue$1.6M

Role in stack

MSAI offers industrial monitoring, sensor software, and data-center reliability context. It is a factory sensing adjacency rather than a clean robot component supplier.

Revenue mix

Q1 2026 revenue was $1.6M, software revenue was $0.7M, gross margin was 57%, adjusted EBITDA loss was $2.1M, and cash was $22.6M.

Proof burden

MSAI needs recurring software conversion, broader customers, lower burn, and clean financing before it should influence the ranked core basket.

MVIS watch lidar portfolio

Lidar and perception portfolio option after asset additions, with revenue bridge, financing, and timing still unresolved.

Market cap$123M
Next earningsNot confirmed
Latest qtr revenue$0.94M

Role in stack

MVIS has lidar and perception optionality through MAVIN, MOVIA, MOSAIK, Luminar assets, and Scantinel assets. It stays watch because commercial proof is weak.

Revenue mix

Q1 2026 revenue was $0.94M, FY2025 revenue was $1.2M, cash was $46.1M, notes were $32.1M, and 2026 guidance was $10M-$15M.

Proof burden

MVIS needs orders that bridge to guidance, positive gross margin, lower cash burn, and reduced ATM/convertible pressure.

LIDR watch pilot-stage lidar

Lowest priority lidar watch: cash and manufacturing capacity are useful, but current revenue and gross profit remain pilot-stage.

Market cap$64M
Next earningsNot confirmed
Latest qtr revenue$0.10M

Role in stack

LIDR is a lidar option with Apollo and LITEON manufacturing context. Capacity helps only if named demand and repeat shipments follow.

Revenue mix

Q1 2026 revenue was $0.101M, gross profit was negative, operating cash flow was negative $8.6M, and cash plus available-for-sale securities were $77.2M.

Proof burden

LIDR needs demand evidence above pilot scale, positive gross margin, burn control, share-count discipline, and named orders that use the LITEON capacity.

Market caps use 2026-07-02 local closes and local weighted shares outstanding. Next earnings dates were not confirmed in the local sources used for this pass. Source-only peers remain Keyence, Omron, SICK, Hikrobot, and private tactile-sensor specialists.

What Confirms Or Weakens

Area What confirms What weakens or invalidates Watch next
01Node thesisSensor content becomes paid deployment
What confirms

Robotics, warehouse, factory, inspection, safety, or industrial automation design wins convert into product revenue, shipped units, repeat orders, software attach, and gross margin.

What weakens or invalidates

Evidence remains automotive-only, prototype-only, or generic sensing language without customer names, volume, price, margin, or repeat shipments.

Watch next
  • Design wins
  • Product revenue
  • Shipped units
  • Customer breadth
02Machine visionCGNX and source-only peers
What confirms

CGNX delivers Q2 guide, holds high gross margin, shows broad order quality, and reports AI vision adoption. Keyence and Omron source-only checks support healthy factory vision demand.

What weakens or invalidates

Orders slow, one customer explains the recovery, margins compress, automation budgets pause, or vision demand stays tied to cyclical factory recovery.

Watch next
  • CGNX Q2
  • AI vision adoption
  • Customer mix
  • Keyence/Omron
033D lidar and localizationHSAI, OUST, AEVA, INVZ
What confirms

HSAI and OUST grow units and product revenue with stable gross margin, better receivable collection, lower burn, and named industrial or robotics customers. AEVA and INVZ convert programs into product revenue.

What weakens or invalidates

ASP cuts, receivable stretch, Section 1260H friction, delayed SOPs, negative gross margin, customer concentration, or equity issuance funds losses before product proof arrives.

Watch next
  • HSAI Q2 guide
  • OUST Q2 guide
  • Receivables
  • Burn rate
04Edge perception computeAMBA conversion
What confirms

AMBA converts CV7/CV8, edge AI, auto, robotics, and IoT design wins into revenue while inventory normalizes and gross margin holds.

What weakens or invalidates

Inventory days stay high, WT/customer concentration worsens, robotics remains unquantified, SBC stays heavy, or GAAP losses fail to narrow.

Watch next
  • Q2 FY2027 revenue
  • Inventory
  • Gross margin
  • Design wins
05Tactile and force sensingLINK and source gaps
What confirms

LINK repeats revenue growth, keeps gross margin above 40%, broadens customers, improves cash flow, and names robotics, end-effector, hand, or force/touch sensor programs.

What weakens or invalidates

Tactile evidence stays at HMI, prototype, private supplier, or marketing level while public revenue remains tiny, concentrated, cash-burning, or thinly traded.

Watch next
  • LINK revenue
  • OCF
  • Customer mix
  • Tactile suppliers
06Specialty sensingCODA, ODYS, MSAI
What confirms

CODA wins DAVD/NANO or Marine orders that map to localization or perception use cases. ODYS and MSAI convert backlog, pilots, or monitoring software into recurring revenue.

What weakens or invalidates

CODA remains mostly subsea/defense niche, Marine weakness persists, ODYS backlog stalls, or MSAI customer concentration and ATM risk dominate.

Watch next
  • CODA orders
  • ODYS backlog
  • MSAI software
  • Customer concentration
07Technical setupTiming remains secondary
What confirms

CGNX, OUST, AMBA, and AEVA hold above short and long local daily trend screens; TDY and LINK keep above long trend; weak names reclaim trend with new operating proof.

What weakens or invalidates

Strong charts lose trend after earnings, HSAI and CODA stay below long trend despite source evidence, or below-trend watch names fail to improve after filings.

Watch next
  • 2026-07-02 OHLC
  • Trend reclaim
  • Volume
  • API charts
08Source qualityRefresh discipline
What confirms

Fresh security lanes, filings, earnings releases, and official disclosures route the claims, while local discovery rows and the report API support chart coverage.

What weakens or invalidates

Source-only peers carry the actual economics, local lanes go stale, the chart API is unavailable, or a new filing changes revenue, margin, cash, guide, customer, or financing facts.

Watch next
  • New filings
  • Lane refresh
  • API status
  • Source gaps

Source Trail

Routes used for this node
Route Use it for Primary links
Paper priorsPerception, tactile sensing, 3D sensors, e-skin, IMUs, multimodal data, VLM/VLA perception, and first-deployment sensing requirements.Citrini humanoid primer; Roland Berger 2026 extraction; Roland Berger PDF
Knowledge posteriorSecurity setup, proof burden, segment evidence, customer concentration, margin, cash conversion, valuation burden, and watch classification.CGNX; HSAI; TDY; OUST; AMBA; CODA; LINK; Information Technology
Watch lanesFunding-sensitive lidar, radar, tactile, visual AI, and monitoring options that need production revenue, gross margin, and financing proof before entering the core.AEVA; INVZ; ARBE; MVIS; LIDR; MSAI; ODYS
Source-only peersMachine-vision, automation, sensing, and tactile/e-skin benchmarks that inform the node but are not ranked as charted securities in this pass.Keyence/KYCCF; Omron/OMRNY; SICK; Hikrobot; private tactile-sensor specialists.
Discovery and DuckDB methodsRead-only local coverage, market-cap calculation, and local daily trend screen. All 14 ranked/watch tickers have daily_ohlc rows through 2026-07-02.node metadata; read-only DuckDB queries on daily_ohlc and instruments; market caps computed from 2026-07-02 close times local weighted shares outstanding.
API chart routeRight-rail chart rendering. Each rank button and metrics strip calls the selected-security panel through data-chart-ticker./api/securities/TICKER/chart?frequency=weekly&window=3y&as_of=latest; no embedded OHLC arrays; no static horizontal chart levels. Local API check on 2026-07-05 returned connection refused at 127.0.0.1:8765.
Ranking processTen xhigh primary subagents and one xhigh merge selector performed the security ranking. The merge kept HSAI above OUST on source-backed lidar scale, kept TDY above OUST on operating quality, kept AMBA core as edge perception compute, demoted CODA to specialty watch, and placed AEVA above LINK among watch names.Final order: CGNX, HSAI, TDY, OUST, AMBA, CODA, AEVA, LINK, INVZ, ARBE, ODYS, MSAI, MVIS, LIDR.
Data limitationsLocal discovery supports price, volume, chart routing, and computed market-cap context. It does not prove humanoid revenue exposure. Papers are priors; filed lanes and official sources are the posterior.Refresh if a new market session, filing, earnings release, guidance update, customer program, financing update, or local API behavior changes.

Categories