Artificial Intelligence

Instructions:

The challenge is a chance for students in Manipal to get a direct entry into the Project MANAS AI task phase. The challenge is open to 1st and 2nd year students. There will be two separate leaderboards for 1st and 2nd years. It is necessary for 2nd year students to solve all of the questions to be considered successful. The hard challenges are optional for 1st year students. The contest will be organized on Hackerrank and will go live at 12:00 AM on 17th September, 2017. The contest will be open till end of September.

Link to the contest: www.hackerrank.com/project-manas-ai-challenge-october-17

 

Taxi Challenge (Easy)

One of the biggest benefits of having driverless cars is easy and convenient taxis around the city. Project MANAS is thinking of launching a taxi service powered by such driverless cars. However, we are having trouble coming up with an optimal algorithm for this.

You are tasked with designing the passenger pickup and drop algorithm for a 2D city grid. Given a list of potential passengers, you need to maximize the profit for the day. You can carry only one passenger.

 

Taxi Challenge (Hard)

One of the biggest benefits of having driverless cars is easy and convenient taxis around the city. Project MANAS is thinking of launching a taxi service powered by such driverless cars. However, we are having trouble coming up with an optimal algorithm for this.

You are tasked with designing the passenger pickup and drop algorithm for a 2D city grid. Given a list of potential passengers, you need to maximize the profit for the day. You can carry multiple passengers.

 

Collateral Damage (Easy)

Project MANAS has designed a new algorithm running on a drone flying above the driverless car. On-board the drone is a low resolution camera, taking satellite view pictures of the region ahead of the car. Each of the two lanes in the road can contain 3 classes of obstacles: Pedestrian, Bike and Car. Each class of obstacle has a corresponding collateral cost value.

Our algorithm takes this image as input and outputs a binary matrix (each cell can either be a ‘0’ or a ‘1’), where a ‘1’ represents that the pixel is classified as road, and a ‘0’ represents an obstacle.

We need your help in designing an algorithm that, given the output of our algorithm, picks the lane with the least collateral damage cost.

 

Collateral Damage (Hard)

Project MANAS has designed a new algorithm running on a drone flying above the driverless car. On-board the drone is a low resolution camera, taking satellite view pictures of the region ahead of the car. Each of the two lanes in the road can contain 3 classes of obstacles: Pedestrian, Bike and Car. Each class of obstacle has a corresponding collateral cost value.

Our algorithm takes this image as input and outputs a binary matrix (each cell can either be a ‘0’ or a ‘1’), where a ‘1’ represents that the pixel is classified as road, and a ‘0’ represents an obstacle.

Unfortunately, our algorithm only has a 95% accuracy in classifying each pixel (i.e. 5% chance that a 0 is actually a 1 or a 1 is actually a 0).

We need your help in designing an algorithm that, given the output of our algorithm, takes into account the erroneous input and picks the lane with the least collateral damage cost.

 

For any queries regarding the challenge, please feel free to contact us at:

 

Sarthak Mittal (Head of Perception)

9810768410

perception_head@projectmanas.in

 

Dheeraj Reddy (Head of Planning)

7406998269

planning_head@projectmanas.in