Dual-steering Robots

 

The iRL is presently using a four-wheeled mobile robot which has steering for both front and rear wheels for conducting its experiments in robotics hardware. We are also researching affordable robots in education in schools and universities with the focus on creating the embedded and teleoperation software for such mobile robots. It is a robust mobile platform on which mobile robots of different functionalities can be implemented. We control it through an API developed in C++. The algorithms we have developed for localization, non-holonomous planning and map-building will be incorporated into the API and tested.

We are currently researching affordable robots for education in schools and universities. In collaboration with our partner Universities, we are focussed on creating the embedded and teleoperation software for such mobile robots

Read more

 

Localization of Mobile Robots

 

Localization is the problem of extracting a most probable position from a series of erroneous and noisy vision and odometry measurements. If the initial position of the robot is known and only compensations for small errors in the robot's odometry is required, the problem is called Local localization. On the contrary, the problem of global localization is where the robot is not informed about its initial position and has to determine the same autonomously. Currently, work on Markov and Monte Carlo methods for global localization have been done. Monte Carlo methods use a sample-based approximation instead of the piecewise constant approximation used in Markov localization. Also, there is work in progress on the use of Kalman filtering for the problem of localization.

 

Planning in high dimensions

 

Planning is the process of determining a trajectory or path for a robot in a given environment. For a simple omni-directional robot, planning is trivial. However, planning a trajectory in the presence of constraints on the motion capabilities (kinematics and dynamics) of a robot requires some amount of computation. Work is underway on popular methods for trajectory planning in two/three dimensions. The methods being used include the Probabilistic Road-Map, and Rapidly-exploring Random search trees.

Read more

 

 

Coverage Maps and Occupancy Grids

 

Generating maps is one of the fundamental tasks of mobile robots. Many successful robotic systems use maps of the environment to perform their tasks. The questions of how to represent environments and how to acquire models using this representation therefore is an active research area. Exploration is the task of guiding a vehicle during mapping such that it covers the environment with its sensors. In addition to the mapping task, efficient exploration strategies are also relevant for surface inspection, mine sweeping, or surveillance.

 

 Read More

Simultaneous Localization and Mapping

The aim is to build a map of the environment based on the laser range data obtained from the robot and at the same time determining where the robot is in the map (localizing). A mobile robot exploring an unknown environment has no absolute frame of reference for its position, other than the features it detects through its sensors. The idea is to solve the object recognition problem and use the distinguishable landmarks to localize the robot.

Read More

Map Building Using Response Grid Method

Here we are generating the map of an unknown area by updating the map cell wise (smallest unit that is updated at a particular instance). Amigobot fires the sonar rays (at certain range and angle) in all directions (covering  whole 360⁰ area ) to detect the obstacles in its surrounding and then the map is updated by marking the covered cells as occupied (having obstacle) or empty. By moving the robot to different positions in the area the whole map is generated.

Read More

Probability based Global Localization

AIM: aim is to obtain the location of the robot in a given map, that is to localize robot globally.
global localization means to estimate the current location of the robot without any prior guess of its initial position. their are two methods to do this:
1 Markov Localization
2 Monte Carlo Method

Read More

Global Localization in dynamic environment

AIM: to identify the location of the robot in a given map where obstacles (for example humans) are moving constantly.
Approach:

Markov method of global localization is implemented where whole map is divided into small size grids and probability of robot being at one of such grids is calculated, in the presence of moving obstacles.

Read More