Rover Profile/ 2023

Encephalon

AI-powered Mars rover with an autonomous onboard planet analyser

Drive: Four-wheeler with custom rocker-bogie suspension
Arm: 6-DOF

CompetitionLEAD // Md. Shaeak Ibna Salim

Scroll

Mission Brief

Mongol-Tori Encephalon is BRAC University's 2023 University Rover Challenge entry, a new-generation Mars rover built to work alongside astronauts and conduct scientific research. This iteration is notable for a fully redesigned science and electrical system, AI-powered subsystems, and significant upgrades across autonomous, network, control, and mechanical systems. The redesigned science module has evolved into an autonomous planet analyser, while a YOLOv5-based AI pipeline classifies tools, equipment, and rocks.Mongol-Tori Encephalon is BRAC University's 2023 University Rover Challenge entry, a new-generation Mars rover built to work alongside astronauts and conduct scientific research. This iteration is notable for a fully redesigned science and electrical system, AI-powered subsystems, and significant upgrades across autonomous, network, control, and mechanical systems. The redesigned science module has evolved into an autonomous planet analyser, while a YOLOv5-based AI pipeline classifies tools, equipment, and rocks.

Spec SheetDWG encephalon-2023-2023

Competition: URC 2023
Year: 2023
Team Lead: Md. Shaeak Ibna Salim

01 / What Makes It New

5 breakthroughs that define Encephalon

01
Innovation
Mongol-Tori Encephalon AI subsystems
An upgraded, AI-powered iteration deploying YOLOv5 models for tool, equipment, and rock classification across retrieval, servicing, and science missions.
02
Innovation
Autonomous planet analyser science module
The fully redesigned, compact science box evolved into an autonomous planet analyser running four soil and atmospheric tests simultaneously via a dedicated science PCB and GUI.
03
Innovation
Dual 5.8 GHz peer-to-peer network
A redesigned network with separately powered IP and analog FPV camera systems on a 5.8 GHz peer-to-peer architecture, using 13 dBi and 19 dBi antennas for over 1 km of coverage.
04
Innovation
Centimeter-accurate RTK autonomy
Dual SparkFun GPS-RTK2 units, a military-grade WitMotion HWT905 IMU, RPLiDar A1, and Intel RealSense D435i deliver centimeter-level waypoint navigation and obstacle avoidance on the Jetson Nano.
05
Innovation
Redesigned EPS and PCB electronics
A custom-PCB electrical system with a dedicated Electric Protection System using ACS758LCB current and B25 voltage sensors, a MOSFET kill switch, and stacked Cytron motor drivers.

02 / Engineering

Built subsystem by subsystem

Every discipline on the team owns a slice of the machine. Here is how each one comes together.

SYS.01Mechanical

01Subsystem

Mechanical

A four-wheeler platform on a triangulated H-shaped ladder chassis with a custom rocker-bogie suspension. New lightweight extender suspension and a tilted-wheel mechanism improve stability during steep vertical drops, while a robust 6-DOF manipulator arm handles retrieval and servicing.

H-shaped ladder chassis 0.7m x 0.55m with triangulation to prevent deformation
Custom rocker-bogie with U-shaped differential bar and two universal joints; bogies turn from 1.2m to 0.76m
New extender suspension powered by a 65 RPM power window motor; tilted-wheel mechanism gives a 60-degree angle for slow descents
Stainless steel wheels (0.29m dia, 0.112m wide) with steel plate rim and 2.5mm rubber grippers
6-DOF arm with chain-and-sprocket wrist, worm gear base lock to prevent backlash, potentiometer feedback per joint, and inverse kinematics
Three-finger claw, an in-development two-finger adaptive claw, and an excavator claw for breaking down rocks and soil
Gazebo for 3D arm simulation and Rviz for arm visualization

SYS.02Network

02Subsystem

Network & Vision

A redesigned dual-subsystem network using a 5.8 GHz peer-to-peer architecture for both IP and analog FPV cameras, powered separately from the main rover. High-gain antennas provide coverage of over 1 km between rover and base station.

Two high-end 5.8 GHz routers in peer-to-peer architecture
Rover: 13 dBi omnidirectional antenna (AMO-5G13) with Rocket 5ac Lite router
Base: 19 dBi 120-degree directional sector antenna (AM-5G19-120) rotating up to 360 degrees, coverage over 1 km
PoE network switch powers two strategically positioned IP cameras
5.8 GHz analog FPV system with 2.1mm adjustable lenses; four analog cameras with 600 mW transmitters/receivers and circular-polarized mushroom antennas
Two FPV cameras accessible simultaneously, others switched via digital switcher with pan mechanism
FPV powered by 12V DC-DC; PoE switch/routers powered by 48V DC-DC and PoE converter

SYS.03Electronics

03Subsystem

Electronics

A redesigned electronics and power distribution system built on custom PCBs, with a dedicated Electric Protection System (EPS) slot. Cytron motor drivers, a relay-based kill switch, and reverse-polarity protection ensure clean power delivery and safety.

Power: 4P4S Li-ion BL18650 (14.6V, 10Ah) for router; Lead Acid (24V, 9Ah) for rest; ~1 hour runtime
Custom PCB with stacked Cytron motor drivers; separate EPS slot
EPS uses ACS758LCB current sensor and B25 voltage sensor, relaying data to base via Arduino Nano
Custom 12V relay kill switch with MOSFET reverse-polarity protection
Dual Channel Cytron 10A DC Motor Driver for arm actuators; Single Channel Cytron 30A Bi-Directional driver for wheels
Manual & debug board for troubleshooting and onboard manual control
Controller board interconnecting Arduino, Raspberry Pi, and Jetson Nano

SYS.04Controls

04Subsystem

Controls & Software

The rover is driven from a portable two-laptop base station running ROS Noetic on Ubuntu 20.04, with Raspberry Pi 4 as the main control architecture. A website-based GUI accessible from any local-network device handles mission control, telemetry, and visualization.

ROS Noetic master node on base station; publisher/subscriber nodes between base and rover
Gamepad and Microsoft Flight Controller for ergonomic maneuvering
ROS bridge plus a robust messaging format with error checking and correction
Website-based GUI with per-mission tabs, joint-angle calculation, speed regulation, voltage and orientation monitoring, 3D rover view, and camera feeds
Lightweight offline map for bird's-eye overview; real-time subsystem dashboard in development
A/B testing of the UI for accessibility
Python socket programming as a supplementary control system independent of ROS

SYS.05Autonomous

05Subsystem

Autonomy

Autonomous navigation runs on an NVIDIA Jetson Nano connected to an Arduino Mega over serial, built on ROS. The navigation node fuses GNSS, IMU, and ArUco tag data to traverse waypoints and detect AR tags, with LiDAR and depth-camera obstacle avoidance.

NVIDIA Jetson Nano processing, Arduino Mega for peripherals via serial
Main navigation node subscribes to GNSS, IMU, and ArUco topics
Military-grade WitMotion HWT905 IMU for heading and compass data
Two SparkFun GPS-RTK2 units (base and rover) for centimeter-level accuracy
RPLiDar A1 for short-range mapping; Intel RealSense D435i depth camera for larger-area path planning
AR-tag detection via OpenCV ArUco library, tag center measured with RealSense depth camera
Custom checkpoint-crossing algorithm; rover rotates 90-degrees per side to find tags; red LED for autonomous start, green LED blink at checkpoints

SYS.06Science

06Subsystem

Science

A fully redesigned, more compact science module functioning as an autonomous planet analyser. It performs four tests across atmospheric and soil analysis, plus AI-based rock classification, with a dedicated science PCB and science-specific GUI.

Redesigned compact science box with distinct compartments and a dedicated science PCB for simultaneous tests
Chemical tests: Ninhydrin (amino acid/protein detection via nichrome-heated tubes) and Iodine (starch/plant cells)
Physical soil analysis: Biomass test (load cell weight difference, (weight difference/previous weight)*100) and water flow capillary test (moisture sensor)
Probe-based NPK test for nitrogen, phosphorus, potassium, plotted in the GUI
Atmospheric sensor module: UV (GUVA-S12SD), DHT22 temp/humidity, MQ2, MQ135, MQ8 (H2), LDR, BMP-280
AI rock classification using YOLOv5 on PyTorch/CNN via microscopic camera texture analysis
Excavator claw and sample collector for soil gathering

03 / Telemetry

The numbers behind the build

Drive System

Four-wheeler with custom rocker-bogie suspension

Chassis

H-shaped ladder chassis, 0.7m x 0.55m

Wheels

Stainless steel, 0.29m dia x 0.112m wide, 2.5mm rubber grippers

Arm DOF

6-DOF arm (360-degree) + 2-DOF end effector

Power

4P4S Li-ion (14.6V, 10Ah) for router; Lead Acid (24V, 9Ah) for rest

Runtime

~1 hour

Comms Range

Over 1 km (5.8 GHz peer-to-peer)

Compute

Raspberry Pi 4, NVIDIA Jetson Nano, Arduino Mega/Nano

Software Stack

ROS Noetic on Ubuntu 20.04 LTS

Parts Index // 30 components

Raspberry Pi 4
NVIDIA Jetson Nano
Arduino Mega
Arduino Nano
ROS Noetic
Ubuntu 20.04 LTS
YOLOv5
PyTorch
OpenCV ArUco
Intel RealSense D435i
RPLiDar A1
SparkFun GPS-RTK2
WitMotion HWT905 IMU
Rocket 5ac Lite router
AMO-5G13 antenna
AM-5G19-120 antenna
Cytron 10A Dual Channel DC Motor Driver
Cytron 30A Bi-Directional DC Motor Driver
ACS758LCB current sensor
B25 voltage sensor
BL18650 Li-ion battery
BMP-280
DHT22
MQ2
MQ135
MQ8
GUVA-S12SD UV sensor
SBG Ellipse-D
Gazebo
Rviz
Raspberry Pi 4
NVIDIA Jetson Nano
Arduino Mega
Arduino Nano
ROS Noetic
Ubuntu 20.04 LTS
YOLOv5
PyTorch
OpenCV ArUco
Intel RealSense D435i
RPLiDar A1
SparkFun GPS-RTK2
WitMotion HWT905 IMU
Rocket 5ac Lite router
AMO-5G13 antenna
AM-5G19-120 antenna
Cytron 10A Dual Channel DC Motor Driver
Cytron 30A Bi-Directional DC Motor Driver
ACS758LCB current sensor
B25 voltage sensor
BL18650 Li-ion battery
BMP-280
DHT22
MQ2
MQ135
MQ8
GUVA-S12SD UV sensor
SBG Ellipse-D
Gazebo
Rviz

Rviz
Gazebo
SBG Ellipse-D
GUVA-S12SD UV sensor
MQ8
MQ135
MQ2
DHT22
BMP-280
BL18650 Li-ion battery
B25 voltage sensor
ACS758LCB current sensor
Cytron 30A Bi-Directional DC Motor Driver
Cytron 10A Dual Channel DC Motor Driver
AM-5G19-120 antenna
AMO-5G13 antenna
Rocket 5ac Lite router
WitMotion HWT905 IMU
SparkFun GPS-RTK2
RPLiDar A1
Intel RealSense D435i
OpenCV ArUco
PyTorch
YOLOv5
Ubuntu 20.04 LTS
ROS Noetic
Arduino Nano
Arduino Mega
NVIDIA Jetson Nano
Raspberry Pi 4
Rviz
Gazebo
SBG Ellipse-D
GUVA-S12SD UV sensor
MQ8
MQ135
MQ2
DHT22
BMP-280
BL18650 Li-ion battery
B25 voltage sensor
ACS758LCB current sensor
Cytron 30A Bi-Directional DC Motor Driver
Cytron 10A Dual Channel DC Motor Driver
AM-5G19-120 antenna
AMO-5G13 antenna
Rocket 5ac Lite router
WitMotion HWT905 IMU
SparkFun GPS-RTK2
RPLiDar A1
Intel RealSense D435i
OpenCV ArUco
PyTorch
YOLOv5
Ubuntu 20.04 LTS
ROS Noetic
Arduino Nano
Arduino Mega
NVIDIA Jetson Nano
Raspberry Pi 4

04 / Mission Plan

Four missions, one machine

How this rover is engineered to score across every University Rover Challenge task.

01
Extreme Retrieval and Delivery Mission
The modified rocker-bogie suspension, expandable body, and rubber-gripper wheels let the rover traverse rocky terrain and travel up to 1 km. A 360-degree three-finger claw lifts heavy equipment with precision, while a YOLOv5 model classifies tools (hammers, toolboxes, screwdrivers) and OpenCV with a sharp distance sensor measures tool distance for efficient pickup.
02
Equipment Servicing Mission
A three-finger claw with inverse-kinematics feedback and a strategically mounted laser pointer give precise control for servicing tasks like toggle switches, joysticks, keyboards, and first aid boxes. Light AI models run on the Jetson Nano for equipment classification and detection.
03
Autonomous Traversal Mission
The rover autonomously navigates between GNSS waypoints using a Jetson Nano fusing dual GPS-RTK2, WitMotion HWT905 IMU, and ArUco tag data. RPLiDar A1 and Intel RealSense D435i provide obstacle detection and shortest-path planning, with a custom checkpoint algorithm and LED status indicators.
04
Science Mission
An autonomous planet analyser runs four tests across atmospheric and soil analysis: Ninhydrin and Iodine chemical tests, Biomass and water-flow physical tests, and a probe-based NPK test, plus AI rock classification. A science-specific GUI and PCB enable simultaneous experiments and data transmission to base.