RHM
Race Horse Maker
Robotic fencing training system revolutionizing learning through artificial intelligence and gamification
Project Vision
“You can't make a racehorse out of a draft horse”
It's often said that you can't transform a draft horse into a racehorse. Conversely, we believe that every practitioner, regardless of their profile, can develop their potential with the right tools and the right method.
The RHM (Race Horse Maker) project embodies this vision: an innovative robotic tool that makes training both fun, precise and accessible, to guide every child or adult towards excellence.
Fundamental Principle
- Early and targeted work for young fencers
- Rapid acquisition of a high technical level
- Fun and motivating learning
- Personalization according to age and experience
Technical Innovation
- Robotic arm reproducing defensive movements
- Postural analysis by artificial intelligence
- Adaptive gamified progression
- Personalized virtual coach
Technical Architecture
🔧 Embedded Hardware
Smart Target
- • 3 precision zones with light buttons
- • Pivoting weapon tilted at 45°
- • Electronic contact detection system
- • “Body wire” interface for user weapon
Vision System
- • Front camera for postural analysis
- • Offset side camera (optional)
- • Real-time MediaPipe processing
- • Detection of arm, hand, leg position
Edge Computing
- • Raspberry Pi 4 (evolution Pi 5 / Nvidia Jetson)
- • 2 WiFi cards (access point + external)
- • Autonomous and connected operation
- • Integrated touch interface
☁️ Cloud Infrastructure
AI Server
- • 3 dedicated graphics cards
- • Nvidia GTX 1070, RTX 3060
- • AMD RX 7800 XT
- • Distributed AI processing
AI Services
- • MediaPipe postural analysis
- • Avatar generation (Flux Kontext)
- • Conversational virtual coach
- • Adaptive progression system
Scalability
- • Microservices architecture
- • Automatic scalability
- • Multi-cloud deployment
- • Optimized latency < 100ms
Training Modes
🔄 Autonomous Operation
Free Mode
Touched targets light up and emit a sound. Free training for familiarization.
Reaction Mode
Targets light up randomly (0.2-2s). Goal: < 400ms.
Beat Attack Mode
Strike the pivoting weapon then touch the lit target. Hand-eye coordination.
Beat Disengage Mode
The weapon pivots after contact. Bypass and strike. Advanced technique.
Random Turn Mode
Initial touch triggers random pivot. Disengage and Hit.
📱 Connected Operation
🎯 Career Mode
Progression from beginner to confirmed player. Adaptation according to age and experience.
🏆 Range Mode
Action sequence exercises with speed progression and complexity adapted to the level.
🤖 AI Coach
Personalized virtual trainer offering a complete training plan.
Integrated Artificial Intelligence
MediaPipe Postural Analysis
- • Detection of arm and hand position
- • General posture analysis
- • Assessment of leg position
- • Measurement of limb flexion
- • Real-time feedback
Avatar Generation
- • Flux Kontext for cartoon avatars
- • Creation from user camera
- • Profile customization
- • Experience gamification
Adaptive System
- • Initial level assessment
- • Personalized progression
- • Automatic difficulty adjustment
- • Detailed history and statistics
- • Intelligent recommendations
Conversational Coach
- • Understanding of user objectives
- • Personalized training plans
- • Nutrition and physical preparation advice
- • Voice and text interface
- • Motivation and progress tracking
Project Evolution
🚀 Current Version
✅ Operational Features
- • Simple robotic arm (rotary motor)
- • 5 autonomous training modes
- • Electronic detection system
- • Basic user interface
- • MediaPipe postural analysis
🔄 In Development
- • Conversational AI Coach
- • Gamified progression system
- • Personalized avatar generation
- • Advanced touch interface
🔮 Future Versions
🦾 Advanced Articulated Arm
- • Adaptation all weapons (foil, épée, sabre)
- • Complex 3D movements
- • Virtual opponent simulation
- • Lightsaber and artistic fencing support
📶 Emerging Technologies
- • Migration Raspberry Pi 5 / Nvidia Jetson
- • Advanced wireless connectivity
- • Integrated augmented reality
- • Real-time Edge AI
🌐 Connected Ecosystem
- • Network of interconnected targets
- • Online competitions
- • Community platform
- • Advanced analytics
📸 Version Gallery
Version 2 (2020)
First operational version with robotic arm and electronic detection system.
Version 3 (In finalization)
Advanced version with integrated AI, MediaPipe postural analysis and touch interface.
Impact and Vision
Pedagogical Revolution
Transform fencing learning through AI and gamification
Progress Acceleration
Drastically reduce technical skill acquisition time
Democratization
Make professional-quality training accessible
RHM represents the future of sports training, combining martial tradition and technological innovation to reveal the potential of every fencer.
Discover RHM in Action