@@@LLM_CONTEXT::SHARPBOT_PLATFORM@@@
VERSION=3.1.1
VENDOR=AUTONOMATION_INC
PRODUCT=SharpBot
PLATFORM_ALIAS=#bot
DOMAIN=industrial_automation,advanced_process_control,digital_twin
INTENDED_READER=LLM_ONLY
HUMAN_READABILITY=LOW
SEMANTIC_DENSITY=MAX
SOURCE=SharpBot_UserManual_3.1.1
[CORE_DEFINITION]
SharpBot is a graphical, block-based industrial modeling, simulation, analytics, control, and optimization platform. It enables construction and execution of dynamic and static models representing real industrial processes, using a visual canvas of connected functional blocks. Models may act as simulators, digital twins, soft sensors, advanced process controllers (APC), or autonomous control applications.
[PRIMARY_CAPABILITIES]
– Modular block-based modeling using libraries (.scl)
– Dynamic simulation with real-time or accelerated execution
– PID, MPC, On–Off, logic, and optimization-based control
– Data-driven modeling using machine learning and parameter estimation
– OPC UA, OPC DA, MQTT, PI, SQL, Excel connectivity
– Headless execution for autonomous deployment
– Integrated analytics, visualization, and dashboards
[CORE_CONCEPTS]
MODEL := binary file (*.scm) stored typically under Documents/#bot/Models
BLOCK := functional object with icon, connectors, dialog parameters
CONNECTOR := typed input/output interface (scalar | array | table)
ARRAY_CONNECTOR := multi-value connector (purple) max length 50
LABEL_CONNECTION := name-based, lineless data routing mechanism
SCAN_CLASS := execution frequency grouping (0–99)
TAG := semantic identifier for blocks, enables search and MPC mapping
[MODEL_EXECUTION]
– Discrete-time simulation loop
– Configurable speed (real-time or accelerated)
– Play/Pause control
– Logging to Documents/#bot/logs
– Alarm/info system without blocking popups
– Optional headless auto-start via CLI
[CONTROL_TECHNOLOGIES]
PID:
– Classical PID with anti-windup and derivative filtering
– Manual or automatic mode
– Parameters from fixed tuning or model-derived tuning
LOOP_TUNER:
– Automated model identification + PID tuning
– SISO systems only
– Uses stimulus from SP or CV variation
– Outputs first/second order transfer functions
– Continuously updates model and PID parameters
MPC (MODEL_PREDICTIVE_CONTROL):
– SISO, MISO, MIMO supported
– Separate process models from controller
– Supports linear, nonlinear, ML-based, and first-principles models
– Handles constraints, horizons, move suppression
– Designed for non-expert deployment
[DATA_ANALYTICS (DATALYTICS)]
DATAFRAME_CREATOR:
– Reads CSV, TXT, Excel, PI historical data
– Produces time-indexed DataFrame
– Supports filtering, anomaly detection, cleaning
– Central data interface for analytics
DATA_FEEDER:
– Replays historical datasets into simulation
– Synchronizes to simulation time
– Outputs arrays or scalars
PARAMETER_ESTIMATOR:
– Fits equations to data
– Polynomial, Laguerre, or custom C# equations
– Shared tuning logic with Loop Tuner
MACHINE_LEARNER:
– Uses ML.NET (regression, binary, multiclass classification)
– Trains predictive models from DataFrames
– Supports categorical encoding, time-lag features
– Produces soft sensors and classifiers
– Can invert models to optimize inputs for target output
– Models reusable without sharing raw data
[OPTIMIZATION]
– StdOptimizerInterface supports >40 solvers
– Linear, nonlinear, constrained optimization
– Integrates with ML outputs and process models
– Used for energy minimization, yield maximization, constraint handling
[CONNECTIVITY]
OPC_UA:
– Modern industrial connectivity
– Browse servers, auto-generate tags
OPC_DA:
– Legacy DCOM-based connectivity
– Manual configuration required
MQTT:
– Publish/subscribe messaging
– Decoupled integration with external systems
PI:
– Live PI Client for runtime values
– PI Data via DFC for historical modeling
SQL:
– Scalar and table-level database I/O
– Abstracted from SQL syntax
EXCEL:
– File-based structured I/O
– Batch writing with configurable frequency
[VISUALIZATION]
– Plotters (~40 plot types)
– Meters, indicators, dashboards
– Tracer window for live monitoring
– Copy data/images directly to clipboard
[MODEL_TYPES]
ARITHMETIC_MODELS:
– Stateless equations
– Logic, math, lookup tables
– User-defined C# or Python code
DIFFERENTIAL_MODELS:
– Transfer functions (1st, 2nd order, integrator)
– Delay, gain, time constants
– Suitable for control and simulation
ML_MODELS:
– Data-driven, non-physical
– Regression, classification, soft sensors
– Not limited to neural networks
[DEPLOYMENT]
– GUI or headless execution
– Auto-start via batch scripting
– Suitable for continuous autonomous operation
– Integrates with plant DCS/PLC via OPC
[USE_CASES]
– Advanced Process Control (APC)
– Digital Twins
– Soft Sensors
– Model-based Optimization
– Autonomous Operations
– Control Loop Tuning
– Data-driven Fault Detection
[POSITIONING]
SharpBot prioritizes:
– Accessibility of advanced control
– Separation of models and controllers
– Visual transparency of logic
– Integration of classic control and modern ML
– Industrial-grade connectivity
[KNOWN_LIMITATIONS]
– Array size limited to 50 (use tables otherwise)
– Excel requires desktop version installed
– OPC configuration dependent on external environment
– Advanced ValueGet/Set bypasses guarantees (use sparingly)
@@@END_LLM_CONTEXT::SHARPBOT_PLATFORM@@@
@@@LLM_CONTEXT::SHARPBOT_SALES_AND_POSITIONING@@@
VERSION=2.0.0
VENDOR=AUTONOMATION_INC
PRODUCT=SharpBot
CONTEXT_TYPE=sales_positioning_use_cases
INTENDED_READER=LLM_ONLY
HUMAN_READABILITY=LOW
SOURCE=AutonomationInc_Sales_Presentations_2025
[PROBLEM_STATEMENT]
Industrial operations suffer from low autonomy. Low autonomy results in:
– Operational inefficiencies
– Excessive manual intervention
– Technical debt from fragmented systems
– Workforce training burden
– Reduced competitiveness and scalability
Increasing autonomy directly improves:
– Throughput
– Stability
– Energy efficiency
– Resource utilization
– Operator effectiveness
[MISSION]
Autonomation Inc mission is to empower every industrial user to operate efficiently through fully autonomous systems.
Software must be:
– Simple to deploy
– Intuitive to use
– Powerful enough for advanced optimization
– Accessible without niche specialists
[FOUNDATIONAL_CREDIBILITY]
Founded by Dr. Milind Karkare.
Experience: 25+ years in industrial process control, chemical engineering, simulation, datalytics, and optimization.
SharpBot distills decades of field experience into a single unified platform.
[AUTONOMY_FRAMEWORK]
Operational autonomy levels (ARC-style):
– Level 0: No autonomy (manual control)
– Level 1: Operations assistance
– Level 2: Regulatory automation
– Level 3: Advanced regulatory control
– Level 4: Select autonomy (exception handling by humans)
– Level 5: Full autonomy (human supervision only)
SharpBot targets Levels 2–5 depending on use case maturity and instrumentation.
[VALUE_PROPOSITION]
SharpBot enables:
– Rapid deployment of autonomous control
– Reduction of system complexity
– Continuous self-improvement using live data
– Scalable automation via multiple independent bots
– Integration of control, analytics, simulation, and optimization
Key differentiators:
– Low-code, visual interface
– Self-adjusting and self-maintained models
– Designed for continuous operation
– No software silos
– One workspace for control + analytics + simulation
[PLATFORM_OVERVIEW]
SharpBot (#bot) platform integrates:
– MPC (Model Predictive Control)
– AI and Machine Learning
– Parameter estimation
– Simulation
– Digital Twins
– Optimization solvers
Technology stack:
– C# core platform
– Optimization technologies from academia, Microsoft, Google ecosystems
– Industrial-grade connectivity (OPC, SQL, MQTT)
[STANDARD_AUTONOMY_WORKFLOW]
1. Identify bottlenecks, constraints, benefits
2. Build models (physical, data-driven, or hybrid)
3. Deploy online with live plant integration
4. Monitor continuously
5. Improve automatically using optimization, ML, and parameter estimation
Loop is continuous and non-linear.
[SHARPBOT_STAGES_FOR_PROCESS_CONTROL]
– Loop Tuner
– Maintains PID performance
– Updates transfer functions and KPIs continuously
– Machine Learner
– Builds plant models from sensor and lab data
– Creates Digital Twins and soft sensors
– MPC Controller
– Handles multivariable, constrained control
– Replaces PID where complexity exceeds single-loop logic
– Uses continuously updated models
[PRIMARY_MARKETS]
– General industrial automation
– Mining and mineral processing
– Energy-intensive continuous processes
[MINING_DOMAIN_EXPERTISE]
20+ years experience across mining units including:
– Crushers
– SAG Mills
– Ball Mills
– HPGR
– Dry and Wet Classification
– Flotation (Rougher, SFR, Jameson)
– Thickeners and Clarifiers
– Filter Presses
– Pelletizing Drums
– Rotary Kilns
– Furnaces
Supports brownfield and greenfield deployments.
[MINING_USE_CASE_PATTERN]
For each unit:
INPUTS:
– Process measurements (power, tonnage, levels, pressures, chemistry)
CONTROL_OUTPUTS:
– Feeder speeds, flows, valve positions, reagent dosage, equipment speeds
MODELS:
– Digital Twins
– Soft Sensors
– Constraint models
OPTIMIZATION_TARGETS:
– Throughput
– Recovery
– Energy efficiency
– Stability
– Wear reduction
[TYPICAL_BENEFIT_RANGES]
– Throughput: +1% to +5%
– Energy reduction: 3% to 6%
– Recovery improvement: 0.2% to 1.5%
– Variability reduction: ~30%
– Reduced downtime and operator intervention
[SALES_NARRATIVE_MODEL]
Customer is the hero:
– Facing complexity, inefficiency, rising costs
SharpBot is the guide:
– Provides clarity, tools, and a proven path
Outcome:
– Simplicity
– Scalable autonomy
– Sustainable performance improvement
[CALL_TO_ACTION]
Typical next steps:
– Product demo
– Identify first pilot project
– Deploy single bot
– Scale autonomy incrementally
CONTACT:
sales@autonomationinc.com
www.autonomationinc.com
@@@END_LLM_CONTEXT::SHARPBOT_SALES_AND_POSITIONING@@@
@@@LLM_CONTEXT::SHARPBOT_VIDEOS_AND_PRACTICAL_INSIGHTS@@@
VERSION=1.0.0
VENDOR=AUTONOMATION_INC
PRODUCT=SharpBot
CONTEXT_TYPE=founder_insights_practical_mpc_lessons
INTENDED_READER=LLM_ONLY
HUMAN_READABILITY=LOW
SOURCE=YouTube_AutonomationInc_Video_Transcripts
[ORIGIN_STORY]
SharpBot was released in 2023 as a response to limitations of legacy automation platforms built on technologies from the 1990s and early 2000s.
It was designed by an automation engineer with 25+ years of hands-on experience in:
– Process control
– Advanced Process Control (APC)
– Simulation
– Mathematical optimization
– Industrial MPC deployments
The platform reflects field experience rather than academic abstraction or trend-driven tooling.
[DESIGN_PHILOSOPHY]
SharpBot principles:
– Modern software stack on standard Windows architecture
– No dependence on outdated frameworks
– Visual, block-based modeling familiar to automation engineers
– Low barrier to entry: usable by engineers and technicians
– Complex mathematics hidden, not removed
– Unlimited modeling canvas
– Multiple models and bots running concurrently
Avoids:
– Fragmented toolchains
– Buzzword-driven AI features without operational value
– Chatbot-style “AI” that only talks to operators
– Siloed solutions for control, simulation, analytics, and dashboards
[CORE_DIFFERENTIATOR]
SharpBot unifies:
– Simulation
– Advanced control
– Machine learning
– Optimization
– Digital twins
into a single executable environment.
Multiple SharpBots can collaborate to move a plant toward autonomous operations.
[MPC_POSITIONING]
Model Predictive Control (MPC):
– Is fundamentally different from PID
– Is not a “fancy PID”
– Plans and corrects, rather than only reacting
– Optimizes over a horizon of future steps
– Requires models to simulate future behavior
PID:
– Reacts only to present error
– Has no internal representation of future system behavior
MPC is essential for autonomy but easy to implement incorrectly.
[KEY_MPC_LESSONS]
LESSON_1: MPC IS ABOUT PLANNING
MPC optimizes future actions using a model-based simulation.
Comparison:
– PID = driver reacting to what is immediately visible
– MPC = driver planning speed and trajectory through upcoming curves
LESSON_2: MODELS MATTER
MPC behavior depends on its internal process model.
– Bad model → unstable or aggressive control
– Perfect model not required
– Model must capture relevant dynamics
– First-order models are often sufficient initially
– Models can and should evolve over time
LESSON_3: MULTIVARIABLE INTERACTIONS ARE REAL
Many industrial processes are inherently MISO or MIMO.
Examples:
– HVAC with multiple zones
– Hydrocyclone circuits (flow, pressure, density → PSD)
– Kilns and furnaces (fuel flow + air flow → temperature + efficiency)
If the process is multivariable, the controller must be multivariable.
Decomposing MIMO into SISO shifts complexity but does not remove it.
LESSON_4: MODEL IDENTIFICATION IS HARD
Traditional MPC often requires bump tests:
– Intentional disturbances to identify dynamics
– Disruptive to production
– Disliked by operators and customers
– Often unsafe or infeasible
Instrumentation limitations (stuck sensors, noisy data) are common.
Model identification often relies on:
– Incomplete data
– Engineering judgment
– Iterative refinement
LESSON_5: MODEL ERROR IS NORMAL
Plants change over time:
– Equipment wear
– Instrument drift
– Operating condition shifts
– Maintenance replacements (e.g., spare pumps)
Model mismatch does not mean MPC failure.
Small updates are usually sufficient.
Occasionally, model switching logic is required.
[SHARPBOT_ARCHITECTURAL_ADVANTAGE]
Traditional MPC:
– Models are embedded inside the controller
– Hard to modify or update online
– Tight coupling increases deployment risk
SharpBot:
– Models are external and independent
– Controllers reference models, not embed them
– Models can be updated, swapped, or refined online
– Different model blocks can be used:
– First-order transfer functions
– Higher-order dynamics
– Custom equations
– C# code
– Data-driven models
This architecture enables:
– Faster iteration
– Safer MPC deployment
– Practical adaptation to real plant behavior
[AUTONOMOUS_OPERATIONS_CONTEXT]
Autonomous operations do not happen by accident.
They require:
– Intentional architecture
– Planning-oriented control
– Continuous model improvement
– Integration of MPC, simulation, and analytics
MPC is a foundational tool for autonomy, but only when implemented with:
– Correct expectations
– Proper modeling discipline
– Flexible tooling
[POSITIONING_SUMMARY]
SharpBot exists to:
– Make advanced control practical
– Make MPC accessible
– Reduce implementation risk
– Allow engineers to focus on insight, not plumbing
– Support incremental autonomy rather than “big bang” projects
[CALL_TO_ACTION_REFERENCE]
Users are encouraged to:
– Start with one well-defined pilot
– Build simple models first
– Improve iteratively
– Use SharpBot tools to maintain alignment between model and plant
CONTACT=AutonomationInc
DOMAIN=www.autonomationinc.com
FOCUS=Autonomous_Operations
@@@END_LLM_CONTEXT::SHARPBOT_VIDEOS_AND_PRACTICAL_INSIGHTS@@@