Stabilization Using Gyroscopic Precession With Real Time Environment Perception
In an environment where control cannot be achieved by chassis movements such as a corner or dead end, some other methods need to be implemented to stabilize the body whilst standing still. Many systems are developed to address this issue i.e. robotic arm to accelerate base but are impractical becaus
2025-06-28 16:36:07 - Adil Khan
Stabilization Using Gyroscopic Precession With Real Time Environment Perception
Project Area of Specialization RoboticsProject SummaryIn an environment where control cannot be achieved by chassis movements such as a corner or dead end, some other methods need to be implemented to stabilize the body whilst standing still. Many systems are developed to address this issue i.e. robotic arm to accelerate base but are impractical because the weight or shape of the active components will always be a hindrance in a practical application. The problem with using mechanical moments to balance a robot is that the more massive the robot chassis is compared to the mass of the balancing mechanism, the less influence the balancing system has on the body. This results in lower controllable range of motion with less maneuverability. Another problem exists in creating a smooth steering system for aircrafts, or rail system for sharper curves without losing the speed.
One such solution to solve all above mentioned problems could be attaching reaction wheel pendulums that’ll achieve stability with the torque exerted by accelerating a relatively weightless flywheel with a clear advantage that it can provide a large amount of torque in a small compact package. Also, it can be significantly efficient because the gyro can spin with a very efficient bearing system using very little power to keep it spinning. This offers a huge advantage over reaction wheel pendulum control used conventionally to balance bodies as a huge proportional amount of power per unit torque is used in the balancing mechanism. Therefore, in many robotics applications where a lightweight active stabilization system is needed that could be possibly done with a precession gyroscope.
Secondly to avoid getting stuck or causing damage to a mobile robot or its surrounding, it should be able to identify obstacles and adapt speed to ground conditions by means of sensors, actuators and control software. By storing properties of the surrounding in the map, a vehicle revisiting an area can benefit from prior information. Similarly for object detection a representation of the geometry of the surrounding can be generated and analyzed in vicinity of the robot and cant be precisely relied on odometery only. Environmental mapping is one of the most important aspects of robotics studies when dealing with localization, positioning, automatic navigation. A real time high resolution 3D mapping Map of unstructured terrain can be generated with use of range sensors and Lidar.The major setback of using radar and sonar range sensor is their low frequency and low precision due to their larger wavelength which won't provide an exact 3D map. In contrast, Lidar will be used which have very high frequency and precision, especially beneficial for short distance measurement. Pan tilt mechanism will be built for the mounting of sensors for creating an exact 3D map with enabled SLAM coded in Matlab and processing IDE.
Project ObjectivesA two wheeled self-balancing stabilizing robot will be developed with smooth non oscillating stabilization achieved through control moment gyroscope. It will not only address the issue of previously balancing technique achieved through large tilting of body and movement of wheels only but will also tackle unexpected abrupt disturbances. Similarly the real hindrance with usage of CMG module of not dealing with continuous disturbance will be solved in this project through change of topography of multiple CMG modules generating high linear torque for greater time period.
A real time perception of the environment will be generated through 3D mapping techniques using LIDAR and range sensors that will create high resolution maps for unstructured terrain making it easier for robot for path planning operations, localization, and teleoperation.Due to data density and high accuracy, LIDAR sensor is very suitable to be implemented in robot mapping and navigation. Hierarchical system design of mapping Lidar will be implemented in a way to be able to model stationary and movable parts of the environment simultaneously and under real-time conditions.The LIDAR-based SLAM algorithm obtains a series of scattered point cloud data with accurate angle and distance information collected by LIDAR, then matches it by ICP (iterative nearest neighbor algorithm), and finally calculates the distance and attitude change of LIDAR relative motion by matching point cloud data at different times to complete the localization of the mobile robot itself. SLAM algorithm will create 3D visualization of the environment processed over to MATLAB GUI for user effectiveness and autonomous navigation through landmark and position state estimate. The greater depth analysis of the map generated will be achieved on Processing IDE with real time analysis.
The behavior based robust controller will be implemented in the robot using H infinity and sliding mode control to eliminate non linearities, errors in switching controllers and zeno phenomenon occurring at boundaries of different obstacles. Control models will be made for convex, concave and hull-shaped obstacles. The algorithms and transfer functions will then be tested in MATLAB and changes will be made accordingly until the desired output of stability is achieved and integrated with aforementioned algorithms to create a two wheeled autonomous robot model with effective environment analysis.
Project Implementation MethodA self adjusting robot will be created like a segway model driven by two actuators consisting of high torque stepper motors, with a joined CMG module on the upper end associated with two gimbal actuators halfway. The CMG module comprises of gimbal and pivoting flywheel which will produce force in one course with explicit load analysis being done. Similarly, a second gimbal will be associated with another end for cancellation of forces and simplicity of turn.
Project begins with 3-D modelling of the drivetrain and CMG assembly. The assemblies are designed in SolidWorks and a motion simulation study is carried out which allows us to verify the dimensions and mass properties of the actual prototype. Using the lagrangian function and mass properties determined from motion simulation we mathematically model the dicycle model for our robot, isolating the CMG module. After simulating the model on MATLAB and designing a PID controller with pole placement for balancing we need to mathematically model our CMG module. Using the mass properties of the assembly from solidWorks we determine the mathematical model isolating any input from the wheels. The robot can then be stabilized in two ways:
- Using input torque from the drive train (stepper motors)
- Using input torque from the CMG module
The CMG module’s primary requirement was to be compact for which we will fabricate a disc shaped flywheel mounted directly on the shaft of a outrunner BLDC motor the gimbal and gimbal actuator assembly will be 3-D printed using ABS plastic as this allows us more freedom in using contours to make the CMG more compact.The drive train assembly will be manufactured using 5mm laser cut acrylic that will be fused together using acrylic glue, as acrylic has greater tensile strength compared to ABS plastic and this technique is also cheaper.
Once the fabrication has been completed designing a smart control algorithm will be initiated. We will be using a behavioural based control algorithm that will employ torque from the wheel and CMG simultaneously to achieve the best possible stability while determining the optimal displacement required by observing the environment around it. H infinity and sliding mode will be implemented and optimized in MATLAB and will prevent zeno phenomenon from occurring within switching behaviour.
A 2D lidar will be used for clear depth analysis of the environment and generating high resolution 3D maps. The LIDAR-based SLAM algorithm obtains a series of scattered point cloud data with accurate angle and distance information collected by LIDAR, matches it by ICP (iterative nearest neighbor algorithm), and finally calculates the distance and attitude change of LIDAR relative motion by matching point cloud data at different times to complete the localization of the mobile robot itself. The servo motors fitted into a pan-tilt mechanism needs to be precisely positioned in order to get accurate point cloud.
Benefits of the ProjectThe foremost benefit of this project is innovation in the field of self-balancing technology. We are achieving an unparalleled stability in the two wheeler realm with an effective steering mechanism. As one of the objectives is to design a compact CMG with high torque to weight ratio the project can be further developed and employed in certain areas where a controlled torque is required in a small package:
-
Allowing vehicles the capability of steering sharper corners and steeper terrains.
-
Integration into bicycles to produce self balancing bicycles that can prevent fatal accidents.
-
A stabilizing backpack for people with trouble walking or disabilities.
-
Generating torque in certain instances of haptic feedback in VR video games.
-
Integrating torque in gameplay controllers.
-
Map a remote area aiding military movements
-
Autonomous navigation and path planning with prior terrain analysis of area too
The final prototype will be a two wheeled robot with a CMG module consisting of dual CMG’s and environment and depth perception capabilities using a camera and LiDAR range finder. The controller for the robot will employ an algorithm that uses torque from the wheels and CMG to stabilise the robot. With environment perception it will be able to judge how ‘confined” the immediate space around it is, this will allow the robot to use the CMG to balance itself without displacing its position. The robot will also be able to use the CMG to negate any sudden disturbances which may directly act upon the robot frame.
A lidar mounted on servo controlled motors with pan tilt mechanism having lower step size for achieving more degree of freedom. A real time analysis and map will be generated with enhanced visualization and depth analysis helpful in navigation and landmarks extraction. Maps will be generated on processing IDE and Matlab. Terrain will be thoroughly analyzed and obstacle collision will be avoided in time. An integrated behaviour with sliding mode behaviour will aid boundary moment in area with a lot of obstacles and goals present and avoid robot getting stuck in an area.
Final Deliverable of the Project HW/SW integrated systemCore Industry TransportationOther Industries Education , Manufacturing , Others Core Technology RoboticsOther Technologies OthersSustainable Development Goals Decent Work and Economic Growth, Industry, Innovation and InfrastructureRequired Resources| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 66400 | |||
| NEMA 17 42mm stepper motor | Equipment | 2 | 900 | 1800 |
| Stepper motor driver TB6560 | Equipment | 2 | 800 | 1600 |
| CMG and chassis fabrication 3D printing | Equipment | 1 | 9000 | 9000 |
| Raspberry Pi 3 B+ | Equipment | 1 | 7000 | 7000 |
| Raspberry Pi Camera | Equipment | 1 | 4500 | 4500 |
| 3S Lipo Battery 5000 mAh | Equipment | 2 | 3500 | 7000 |
| Camera Mount with Servo Motors | Equipment | 1 | 1500 | 1500 |
| LiDAR range finder sensor module | Equipment | 1 | 18000 | 18000 |
| Attitude sensor GY-BNO055 | Equipment | 1 | 3500 | 3500 |
| Iron flywheel | Equipment | 2 | 1000 | 2000 |
| Outrunner BLDC motor - 930KV | Equipment | 2 | 700 | 1400 |
| 3S 20A bi-directional brushless motor ESC | Equipment | 2 | 1500 | 3000 |
| STM-32f103C8 microcontroller | Equipment | 1 | 600 | 600 |
| Gimbal actuator motors | Equipment | 2 | 750 | 1500 |
| Wires, boards, power supply, printing, glue | Miscellaneous | 1 | 4000 | 4000 |