Remove all previously tracked files that are now ignored

.gitignore

@@ -1,2 +1,3 @@
 .env
+__pycache__/
 *.pyc

ai_intelligence_layer/.env

@@ -1,25 +0,0 @@
-# Gemini API Configuration
-GEMINI_API_KEY=AIzaSyDK_jxVlJUpzyxuiGcopSFkiqMAUD3-w0I
-GEMINI_MODEL=gemini-2.5-flash
-
-# Service Configuration
-AI_SERVICE_PORT=9000
-AI_SERVICE_HOST=0.0.0.0
-
-# Enrichment Service Integration
-ENRICHMENT_SERVICE_URL=http://localhost:8000
-ENRICHMENT_FETCH_LIMIT=10
-
-# Demo Mode (enables caching and consistent responses for demos)
-DEMO_MODE=false
-
-# Fast Mode (use shorter prompts for faster responses)
-FAST_MODE=true
-
-# Strategy Generation Settings
-STRATEGY_COUNT=3  # Number of strategies to generate (3 for testing, 20 for production)
-
-# Performance Settings
-BRAINSTORM_TIMEOUT=90
-ANALYZE_TIMEOUT=120
-GEMINI_MAX_RETRIES=3

ai_intelligence_layer/main.py

@@ -49,6 +49,7 @@ strategy_generator: StrategyGenerator = None
 telemetry_client: TelemetryClient = None
 current_race_context: RaceContext = None  # Store race context globally
 last_control_command: Dict[str, int] = {"brake_bias": 5, "differential_slip": 5}  # Store last command
+strategy_history: List[Dict[str, Any]] = []  # Track past strategies for continuity
 
 # WebSocket connection manager
 class ConnectionManager:
@@ -378,7 +379,7 @@ async def websocket_pi_endpoint(websocket: WebSocket):
     2. AI layer processes telemetry and generates strategies
     3. AI layer pushes control commands back to Pi (brake_bias, differential_slip)
     """
-    global current_race_context, last_control_command
+    global current_race_context, last_control_command, strategy_history
     
     vehicle_id = await websocket_manager.connect(websocket)
     
@@ -389,7 +390,11 @@ async def websocket_pi_endpoint(websocket: WebSocket):
     # Reset last control command to neutral for new session
     last_control_command = {"brake_bias": 5, "differential_slip": 5}
     
+    # Clear strategy history for new race
+    strategy_history = []
+    
     logger.info("[WebSocket] Telemetry buffer cleared for new connection")
+    logger.info("[WebSocket] Strategy history cleared for new race")
     
     # Notify dashboards of new vehicle connection
     await dashboard_manager.broadcast({
@@ -454,12 +459,26 @@ async def websocket_pi_endpoint(websocket: WebSocket):
                             try:
                                 response = await strategy_generator.generate(
                                     enriched_telemetry=buffer_data,
-                                    race_context=current_race_context
+                                    race_context=current_race_context,
+                                    strategy_history=strategy_history
                                 )
                                 
                                 # Extract top strategy (first one)
                                 top_strategy = response.strategies[0] if response.strategies else None
                                 
+                                # Add to strategy history
+                                if top_strategy:
+                                    strategy_history.append({
+                                        "lap": lap_number,
+                                        "strategy_name": top_strategy.strategy_name,
+                                        "risk_level": top_strategy.risk_level,
+                                        "brief_description": top_strategy.brief_description,
+                                        "reasoning": top_strategy.reasoning
+                                    })
+                                    # Keep only last 10 strategies
+                                    if len(strategy_history) > 10:
+                                        strategy_history.pop(0)
+                                
                                 # Generate control commands based on strategy
                                 control_command = generate_control_command(
                                     lap_number=lap_number,
@@ -490,6 +509,11 @@ async def websocket_pi_endpoint(websocket: WebSocket):
                                     "type": "lap_data",
                                     "vehicle_id": vehicle_id,
                                     "lap_data": enriched,
+                                    "race_context": {
+                                        "position": current_race_context.driver_state.current_position,
+                                        "gap_to_leader": current_race_context.driver_state.gap_to_leader,
+                                        "gap_to_ahead": current_race_context.driver_state.gap_to_ahead
+                                    },
                                     "control_output": {
                                         "brake_bias": control_command["brake_bias"],
                                         "differential_slip": control_command["differential_slip"]
@@ -497,7 +521,8 @@ async def websocket_pi_endpoint(websocket: WebSocket):
                                     "strategy": {
                                         "strategy_name": top_strategy.strategy_name,
                                         "risk_level": top_strategy.risk_level,
-                                        "brief_description": top_strategy.brief_description
+                                        "brief_description": top_strategy.brief_description,
+                                        "reasoning": top_strategy.reasoning
                                     } if top_strategy else None,
                                     "timestamp": datetime.now().isoformat()
                                 })
@@ -527,6 +552,11 @@ async def websocket_pi_endpoint(websocket: WebSocket):
                                 "type": "lap_data",
                                 "vehicle_id": vehicle_id,
                                 "lap_data": enriched,
+                                "race_context": {
+                                    "position": current_race_context.driver_state.current_position,
+                                    "gap_to_leader": current_race_context.driver_state.gap_to_leader,
+                                    "gap_to_ahead": current_race_context.driver_state.gap_to_ahead
+                                },
                                 "control_output": {
                                     "brake_bias": 5,
                                     "differential_slip": 5

ai_intelligence_layer/models/input_models.py

@@ -62,6 +62,7 @@ class Strategy(BaseModel):
     pit_laps: List[int] = Field(..., description="Lap numbers for pit stops")
     tire_sequence: List[Literal["soft", "medium", "hard", "intermediate", "wet"]] = Field(..., description="Tire compounds in order")
     brief_description: str = Field(..., description="One sentence rationale")
+    reasoning: Optional[str] = Field(None, description="Detailed explanation including strategy change/continuity rationale")
     risk_level: Literal["low", "medium", "high", "critical"] = Field(..., description="Risk assessment")
     key_assumption: str = Field(..., description="Main assumption this strategy relies on")
 

ai_intelligence_layer/prompts/brainstorm_prompt.py

@@ -9,7 +9,8 @@ from config import get_settings
 
 def build_brainstorm_prompt_fast(
     enriched_telemetry: List[EnrichedTelemetryWebhook],
-    race_context: RaceContext
+    race_context: RaceContext,
+    strategy_history: List[dict] = None
 ) -> str:
     """Build a faster, more concise prompt for quicker responses (lap-level data)."""
     settings = get_settings()
@@ -27,17 +28,29 @@ def build_brainstorm_prompt_fast(
     if gap_to_leader > 0 and position > 1:
         comp_info += f", {gap_to_leader:.1f}s from leader"
     
+    # Format strategy history (last 3 strategies)
+    history_text = ""
+    if strategy_history and len(strategy_history) > 0:
+        recent_history = strategy_history[-3:]  # Last 3 strategies
+        history_lines = []
+        for h in recent_history:
+            history_lines.append(f"Lap {h['lap']}: {h['strategy_name']} (Risk: {h['risk_level']})")
+        history_text = f"\n\nPAST STRATEGIES:\n" + "\n".join(history_lines)
+        history_text += f"\n\nREQUIREMENT: If changing from previous strategy '{recent_history[-1]['strategy_name']}', explain WHY the switch is necessary. If staying with same approach, explain continuity."
+    else:
+        history_text = "\n\nNOTE: This is the first strategy generation - no previous strategy to compare."
+    
     if count == 1:
         # Ultra-fast mode: just generate 1 strategy
         return f"""Generate 1 F1 race strategy for {race_context.driver_state.driver_name} at {race_context.race_info.track_name}.
 
 CURRENT: Lap {race_context.race_info.current_lap}/{race_context.race_info.total_laps}, {comp_info}, {race_context.driver_state.current_tire_compound} tires ({race_context.driver_state.tire_age_laps} laps old)
 
-TELEMETRY: Tire deg {latest.tire_degradation_rate:.2f}, Cliff risk {latest.tire_cliff_risk:.2f}, Pace {latest.pace_trend}, Position trend {latest.position_trend}, Competitive pressure {latest.competitive_pressure:.2f}
+TELEMETRY: Tire deg {latest.tire_degradation_rate:.2f}, Cliff risk {latest.tire_cliff_risk:.2f}, Pace {latest.pace_trend}, Position trend {latest.position_trend}, Competitive pressure {latest.competitive_pressure:.2f}{history_text}
 
 Generate 1 optimal strategy considering competitive position. Min 2 tire compounds required.
 
-JSON: {{"strategies": [{{"strategy_id": 1, "strategy_name": "name", "stop_count": 1, "pit_laps": [32], "tire_sequence": ["medium", "hard"], "brief_description": "one sentence", "risk_level": "medium", "key_assumption": "main assumption"}}]}}"""
+JSON: {{"strategies": [{{"strategy_id": 1, "strategy_name": "name", "stop_count": 1, "pit_laps": [32], "tire_sequence": ["medium", "hard"], "brief_description": "one sentence", "reasoning": "detailed explanation including strategy change rationale if applicable", "risk_level": "medium", "key_assumption": "main assumption"}}]}}"""
     
     elif count <= 5:
         # Fast mode: 2-5 strategies with different approaches
@@ -47,11 +60,13 @@ CURRENT: Lap {race_context.race_info.current_lap}/{race_context.race_info.total_
 
 TELEMETRY: Tire deg {latest.tire_degradation_rate:.2f}, Cliff risk {latest.tire_cliff_risk:.2f}, Pace {latest.pace_trend}, Delta {latest.performance_delta:+.2f}s, Position trend {latest.position_trend}, Competitive pressure {latest.competitive_pressure:.2f}
 
-COMPETITIVE SITUATION: Gap to ahead {gap_to_ahead:.1f}s ({"DRS RANGE - attack opportunity!" if gap_to_ahead < 1.0 else "close battle" if gap_to_ahead < 3.0 else "need to push"})
+COMPETITIVE SITUATION: Gap to ahead {gap_to_ahead:.1f}s ({"DRS RANGE - attack opportunity!" if gap_to_ahead < 1.0 else "close battle" if gap_to_ahead < 3.0 else "need to push"}){history_text}
 
 Generate {count} strategies balancing tire management with competitive pressure. Consider if aggressive undercut makes sense given gaps. Min 2 tire compounds each.
 
-JSON: {{"strategies": [{{"strategy_id": 1, "strategy_name": "Conservative Stay Out", "stop_count": 1, "pit_laps": [35], "tire_sequence": ["medium", "hard"], "brief_description": "extend current stint then hard tires to end", "risk_level": "low", "key_assumption": "tire cliff risk stays below 0.7"}}]}}"""
+CRITICAL: Each strategy MUST include 'reasoning' field explaining the approach and, if applicable, why it differs from or continues the previous strategy.
+
+JSON: {{"strategies": [{{"strategy_id": 1, "strategy_name": "Conservative Stay Out", "stop_count": 1, "pit_laps": [35], "tire_sequence": ["medium", "hard"], "brief_description": "extend current stint then hard tires to end", "reasoning": "Detailed explanation of this strategy including why we're switching from/continuing previous approach based on current telemetry and competitive situation", "risk_level": "low", "key_assumption": "tire cliff risk stays below 0.7"}}]}}"""
     
     return f"""Generate {count} F1 race strategies for {race_context.driver_state.driver_name} at {race_context.race_info.track_name}.
 
@@ -59,16 +74,19 @@ CURRENT: Lap {race_context.race_info.current_lap}/{race_context.race_info.total_
 
 TELEMETRY: Tire deg {latest.tire_degradation_rate:.2f}, Cliff risk {latest.tire_cliff_risk:.2f}, Pace {latest.pace_trend}, Delta {latest.performance_delta:+.2f}s, Position trend {latest.position_trend}, Competitive pressure {latest.competitive_pressure:.2f}
 
-COMPETITIVE: Gap ahead {gap_to_ahead:.1f}s, Position trending {latest.position_trend}
+COMPETITIVE: Gap ahead {gap_to_ahead:.1f}s, Position trending {latest.position_trend}{history_text}
 
 Generate {count} diverse strategies considering both tire management AND competitive positioning. Min 2 compounds.
 
-JSON: {{"strategies": [{{"strategy_id": 1, "strategy_name": "name", "stop_count": 1, "pit_laps": [32], "tire_sequence": ["medium", "hard"], "brief_description": "one sentence", "risk_level": "low|medium|high|critical", "key_assumption": "main assumption"}}]}}"""
+CRITICAL: Each strategy MUST include 'reasoning' field with detailed rationale and strategy continuity/change explanation.
+
+JSON: {{"strategies": [{{"strategy_id": 1, "strategy_name": "name", "stop_count": 1, "pit_laps": [32], "tire_sequence": ["medium", "hard"], "brief_description": "one sentence", "reasoning": "Comprehensive explanation including strategy evolution context", "risk_level": "low|medium|high|critical", "key_assumption": "main assumption"}}]}}"""
 
 
 def build_brainstorm_prompt(
     enriched_telemetry: List[EnrichedTelemetryWebhook],
-    race_context: RaceContext
+    race_context: RaceContext,
+    strategy_history: List[dict] = None
 ) -> str:
     """
     Build the brainstorm prompt for Gemini.
@@ -76,6 +94,7 @@ def build_brainstorm_prompt(
     Args:
         enriched_telemetry: Recent enriched telemetry data
         race_context: Current race context
+        strategy_history: List of previous strategies for context
         
     Returns:
         Formatted prompt string
@@ -108,6 +127,14 @@ def build_brainstorm_prompt(
             "gap_seconds": round(c.gap_seconds, 1)
         })
     
+    # Format strategy history
+    history_section = ""
+    if strategy_history and len(strategy_history) > 0:
+        history_section = "\n\nSTRATEGY HISTORY (Recent decisions):\n"
+        for h in strategy_history[-5:]:  # Last 5 strategies
+            history_section += f"- Lap {h['lap']}: {h['strategy_name']} (Risk: {h['risk_level']}) - {h.get('brief_description', 'N/A')}\n"
+        history_section += f"\nCONTEXT: Previous strategy was '{strategy_history[-1]['strategy_name']}'. If recommending a change, clearly explain WHY. If continuing same approach, explain the continuity.\n"
+    
     prompt = f"""You are an expert F1 strategist. Generate 20 diverse race strategies based on lap-level telemetry AND competitive positioning.
 
 LAP-LEVEL TELEMETRY METRICS:
@@ -135,7 +162,7 @@ COMPETITORS:
 {competitors_data}
 
 ENRICHED TELEMETRY (Last {len(telemetry_data)} laps, newest first):
-{telemetry_data}
+{telemetry_data}{history_section}
 
 KEY INSIGHTS:
 - Latest tire degradation rate: {latest.tire_degradation_rate:.3f}
@@ -160,9 +187,12 @@ For each strategy provide:
 - pit_laps: [array of lap numbers]
 - tire_sequence: [array of compounds: "soft", "medium", "hard"]
 - brief_description: One sentence rationale
+- reasoning: DETAILED explanation (3-5 sentences) including: (1) Why this strategy fits current conditions, (2) How it addresses telemetry trends, (3) Strategy evolution - why changing from or continuing previous approach
 - risk_level: "low", "medium", "high", or "critical"
 - key_assumption: Main assumption this strategy relies on
 
+CRITICAL: The 'reasoning' field must include strategy continuity/change rationale relative to past decisions.
+
 OUTPUT FORMAT (JSON only, no markdown):
 {{
   "strategies": [
@@ -173,6 +203,7 @@ OUTPUT FORMAT (JSON only, no markdown):
       "pit_laps": [32],
       "tire_sequence": ["medium", "hard"],
       "brief_description": "Extend mediums to lap 32, safe finish on hards",
+      "reasoning": "Current tire degradation at 0.45 suggests we can safely extend to lap 32 before hitting the cliff. This maintains our conservative approach from lap 3 as conditions haven't changed significantly - tire deg is stable and we're not under immediate competitive pressure. The hard tire finish provides low-risk race completion.",
       "risk_level": "low",
       "key_assumption": "Tire degradation stays below 0.85 until lap 32"
     }}

ai_intelligence_layer/services/strategy_generator.py

@@ -25,7 +25,8 @@ class StrategyGenerator:
     async def generate(
         self,
         enriched_telemetry: List[EnrichedTelemetryWebhook],
-        race_context: RaceContext
+        race_context: RaceContext,
+        strategy_history: List[dict] = None
     ) -> BrainstormResponse:
         """
         Generate 20 diverse race strategies.
@@ -33,6 +34,7 @@ class StrategyGenerator:
         Args:
             enriched_telemetry: Recent enriched telemetry data
             race_context: Current race context
+            strategy_history: List of previous strategies for continuity
             
         Returns:
             BrainstormResponse with 20 strategies
@@ -41,13 +43,15 @@ class StrategyGenerator:
             Exception: If generation fails
         """
         logger.info(f"Generating strategies using {len(enriched_telemetry)} laps of telemetry")
+        if strategy_history:
+            logger.info(f"Including {len(strategy_history)} previous strategies in context")
         
         # Build prompt (use fast mode if enabled)
         if self.settings.fast_mode:
             from prompts.brainstorm_prompt import build_brainstorm_prompt_fast
-            prompt = build_brainstorm_prompt_fast(enriched_telemetry, race_context)
+            prompt = build_brainstorm_prompt_fast(enriched_telemetry, race_context, strategy_history or [])
         else:
-            prompt = build_brainstorm_prompt(enriched_telemetry, race_context)
+            prompt = build_brainstorm_prompt(enriched_telemetry, race_context, strategy_history or [])
         
         # Generate with Gemini (high temperature for creativity)
         response_data = await self.gemini_client.generate_json(

ai_intelligence_layer/static/dashboard.html

@@ -715,14 +715,14 @@
         }
         
         function handleMessage(data) {
-            const { type, vehicle_id, lap_data, control_output, strategy, timestamp } = data;
+            const { type, vehicle_id, lap_data, race_context, control_output, strategy, timestamp } = data;
             
             if (type === 'vehicle_connected') {
                 addVehicle(vehicle_id, timestamp);
             } else if (type === 'vehicle_disconnected') {
                 removeVehicle(vehicle_id);
             } else if (type === 'lap_data') {
-                addLapData(vehicle_id, lap_data, control_output, strategy, timestamp);
+                addLapData(vehicle_id, lap_data, race_context, control_output, strategy, timestamp);
             }
         }
         
@@ -750,7 +750,7 @@
             }
         }
         
-        function addLapData(vehicleId, lapData, controlOutput, strategy, timestamp) {
+        function addLapData(vehicleId, lapData, raceContext, controlOutput, strategy, timestamp) {
             if (!vehicles.has(vehicleId)) {
                 addVehicle(vehicleId, timestamp);
             }
@@ -758,6 +758,7 @@
             const vehicle = vehicles.get(vehicleId);
             vehicle.laps.push({
                 ...lapData,
+                race_context: raceContext,  // Add race context with position and gaps
                 control_output: controlOutput,
                 strategy: strategy,
                 timestamp: timestamp
@@ -907,6 +908,37 @@
             
             let bodyHtml = '';
             
+            // Race Position section (lap_time, position, gaps)
+            if (lap.race_context || lap.lap_time) {
+                bodyHtml += `
+                    <div class="modal-section">
+                        <h3>Race Position</h3>
+                        <div class="modal-data">
+                            ${lap.lap_time ? `
+                            <div class="modal-data-row">
+                                <span class="modal-data-label">Lap Time</span>
+                                <span class="modal-data-value">${lap.lap_time}</span>
+                            </div>
+                            ` : ''}
+                            ${lap.race_context ? `
+                            <div class="modal-data-row">
+                                <span class="modal-data-label">Position</span>
+                                <span class="modal-data-value">P${lap.race_context.position}</span>
+                            </div>
+                            <div class="modal-data-row">
+                                <span class="modal-data-label">Gap from Leader</span>
+                                <span class="modal-data-value">${lap.race_context.gap_to_leader > 0 ? '+' + lap.race_context.gap_to_leader.toFixed(3) + 's' : (lap.race_context.position === 1 ? 'Leading' : 'N/A')}</span>
+                            </div>
+                            <div class="modal-data-row">
+                                <span class="modal-data-label">Gap from Car Ahead</span>
+                                <span class="modal-data-value">${lap.race_context.gap_to_ahead > 0 ? '+' + lap.race_context.gap_to_ahead.toFixed(3) + 's' : (lap.race_context.position === 1 ? 'Leading' : 'N/A')}</span>
+                            </div>
+                            ` : ''}
+                        </div>
+                    </div>
+                `;
+            }
+            
             // Telemetry data section
             bodyHtml += `
                 <div class="modal-section">