New opportunity

Research Fellow (Embodied AI)

Nanyang Technological University · Singapore

About this role

Nanyang Technological University’s School of Mechanical & Aerospace Engineering seeks a Research Fellow to advance embodied AI and robotic manipulation. The research focuses on AI-driven robotic systems for real-world applications, including human-robot interaction, robotic perception and intelligent control.

Responsibilities

Develop and implement perception and control algorithms for robotic arms and embodied AI systems, and integrate them into robotic platforms.
Help integrate multimodal AI models using vision, language and force-sensor inputs with robotic systems.
Test AI-driven robotic behavior through experiments in simulated and real-world environments, working with robotic manipulators and mobile robots.
Contribute to research publications, reports and presentations; collaborate with faculty, postdoctoral researchers and industry partners; and help maintain and operate laboratory equipment and robotic systems.

Required qualifications

A PhD in Computer Science, Robotics, Mechanical Engineering, Electrical Engineering or a related field, with at least three first-author papers.
Experience in robotic perception, control or learning-based methods; proficiency in Python and C++ for AI and robotics; and familiarity with robotic middleware such as ROS or MoveIt and with hardware integration.

Preferred qualifications

Knowledge of machine learning, reinforcement learning or vision-language models for robotics is a plus. Hands-on experience with robotic arms such as UR5 or Franka Emika is desirable.

This is a full-time position at NTU Main Campus, Singapore.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

Embodied AIRobotic manipulationRobotic perceptionRobot controlMultimodal AIVision-language modelsMachine learningReinforcement learningPythonC++ROSMoveItHardware integrationRobotic armsMobile robotsUR5Franka EmikaResearch publicationCollaboration

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Perception and control for manipulation
The posting calls for developing and implementing algorithms for robotic arms and embodied AI systems, then integrating them into robotic platforms. Python, C++, robotic middleware and hardware integration are stated requirements.
Multimodal robotic integration
The role assists with connecting vision, language and force-sensor inputs to robotic systems. Knowledge of machine learning, reinforcement learning or vision-language models for robotics is listed as a plus, not a requirement.
Simulation and real-world experimentation
The fellow will test AI-driven behavior in simulated and real-world environments, work with manipulators and mobile robots, and help operate and maintain lab equipment.
Research output and collaboration
The role supports publications, reports and presentations and works with faculty, postdocs and industry partners. The posting also requires a PhD and at least three first-author papers in a listed or related field.

Likely challenges to think through

Transferring results to hardware
Preparation hypothesis: Behavior that works in simulation may be harder to reproduce on physical robots, where reliability must be balanced against experimental speed. Prepare an example of how you would validate an algorithm across both settings and investigate a mismatch.
Combining different sensor inputs
Preparation hypothesis: Vision, language and force inputs may differ in timing and reliability, complicating control decisions. Prepare to explain how you would test an integration, handle uncertain inputs and determine whether it improves robotic behavior.
Balancing novelty with usable results
Preparation hypothesis: Novel algorithms must also support platform experiments and research outputs, creating a trade-off between exploratory work and reproducible evidence. Prepare to describe how you would scope experiments, record results and communicate limitations to collaborators.

Practice your answers

Technical practice

How would you design a perception pipeline for a robotic arm performing manipulation?

What to coverThe role includes developing robotic perception algorithms. Preparation suggestion: Explain how you would represent objects and their poses, handle uncertain observations, and check whether perception is accurate enough to guide manipulation.

How would you choose and evaluate a control approach for a robotic arm?

What to coverThe role includes developing and implementing control algorithms. Preparation suggestion: Discuss the trade-offs between tracking accuracy, responsiveness, robustness, and safe behavior when contact or uncertainty is involved.

How would you combine vision, language, and force-sensor inputs in a robotic system?

What to coverThe posting explicitly mentions integrating these modalities. Preparation suggestion: Explain what each input contributes, how you would align them with robot actions, and how the system should respond when the inputs disagree.

How would you test whether behavior learned or developed in simulation works on a real robot?

What to coverThe role calls for experiments in simulated and real-world environments. Preparation suggestion: Describe a staged evaluation, identify likely gaps between simulation and hardware, and explain how test results would inform revisions.

What evidence would show that an AI-driven manipulation behavior is reliable?

What to coverTesting robotic behavior is a stated responsibility. Preparation suggestion: Define task-relevant success and failure measures, vary conditions across trials, and distinguish repeatable performance from an isolated successful demonstration.

How would you integrate a perception or control component using ROS or MoveIt?

What to coverThe posting asks for familiarity with robotic middleware such as ROS or MoveIt and hardware integration. Preparation suggestion: Describe component interfaces, how you would verify data and command flow, and where integration failures might arise.

How would you divide robotics software work between Python and C++?

What to coverProficiency in both languages for AI and robotics applications is listed. Preparation suggestion: Explain trade-offs in iteration speed, runtime performance, and maintainability, using a robotics workflow rather than assuming a particular codebase.

When would a learning-based approach be preferable to another approach for robotic manipulation?

What to coverThe role accepts experience in learning-based methods; machine learning and reinforcement learning knowledge is a plus. Preparation suggestion: Compare data needs, generalization, interpretability, and evaluation burden without treating reinforcement learning as mandatory.

How would you diagnose inconsistent behavior on a robotic manipulator or mobile robot?

What to coverThe role involves hardware platforms and operating robotic systems. Preparation suggestion: Describe how you would separate sensor, software, control, and hardware causes, then test changes without obscuring the original failure.

How would you frame a novel embodied-AI algorithm as a research contribution?

What to coverThe role involves developing algorithms and supporting publications, reports, and presentations. Preparation suggestion: State the research question, relevant comparisons, experimental evidence, and limitations needed to substantiate a contribution.

Behavioural practice

Tell me about a time you collaborated with researchers from different specialties on a robotics project.

What to coverThe posting calls for collaboration with faculty and postdoctoral researchers. Preparation suggestion: Use a real situation, your actions to coordinate technical decisions, and the result; clarify your own contribution.

Tell me about a time you worked with an external partner whose practical priorities differed from research goals.

What to coverCollaboration with industry partners and real-world applications are part of the role. Preparation suggestion: Use a real situation, what you did to reconcile priorities, and the result, without assuming a specific partner arrangement at NTU.

Tell me about an experiment that did not support your initial approach.

What to coverThe role includes testing robotic behavior in simulation and on hardware. Preparation suggestion: Explain the situation, how you investigated or revised the approach, and the resulting evidence or lesson.

Tell me about your contribution to a research publication or presentation.

What to coverThe posting includes research outputs and requires at least three first-author papers. Preparation suggestion: Choose a real example and describe the research context, your specific actions, and the outcome, distinguishing your contribution from the team's.

Tell me about a time you helped keep shared robotics equipment usable for others.

What to coverAssisting with laboratory equipment and robotic-system operation is a stated duty. Preparation suggestion: Give a real situation, the actions you took to address an operational issue or coordinate use, and the result for the team.

Sharpen your knowledge

Topics to prepare

Robotic perception and intelligent control
Revise how perception estimates the state needed for robotic-arm control, and trace how estimation errors affect a manipulation task. The posting explicitly calls for perception and control algorithms and experience in either area or learning-based methods.
Embodied AI for manipulation
Sketch the decision loop from sensing to action for a robotic arm, then explain where a learned method could replace or complement a control component. Embodied AI and robotic manipulation are the stated research focus; the particular approach is a preparation suggestion, not a specified requirement.
Multimodal sensing for robotics
Draw a conceptual flow showing how vision, language and force-sensor inputs could jointly guide a manipulation action; identify a potential conflict between inputs and how to handle it. Integrating these modalities is an explicit responsibility.
Evaluating simulated and real-world behavior
Outline an experiment that runs the same robotic behavior in simulation and on hardware, with a measurable outcome and a discussion of discrepancies. The role explicitly involves experiments in both environments with manipulators and mobile robots.
Research contributions and evidence
Revisit a first-author paper and practice stating its research question, novel contribution, experimental evidence and limitations in two minutes. The posting requires a PhD and at least three first-author papers, and includes publications, reports and presentations.

Skills to revise

Implement algorithms in Python and C++
As a revision exercise, implement a small perception-to-control routine in each language and check that both produce consistent outputs on the same test inputs. Proficiency in Python and C++ for AI and robotics is explicitly required.
Work through ROS or MoveIt integration
Review a robotics integration example using ROS or MoveIt and trace the path from an input or command to the robot interface, noting likely points of failure. The posting requires familiarity with robotic middleware such as these and with hardware integration.
Integrate a robotic arm with sensors
Walk through a practical checklist for connecting an arm, checking sensor readings and verifying commanded versus observed motion before a manipulation experiment. Hardware-platform work is an explicit duty; UR5 and Franka Emika experience is desirable, not required.
Test and document robot experiments
Draft a compact test plan with a task, repeatable conditions, behavior measures and a way to record failures in simulation and on physical hardware. This prepares for the stated responsibility to test AI-driven robotic behavior in both settings.
Present research for collaborators
Prepare a short explanation of one research result with its method, evidence and next experiment, tailored for faculty, postdocs or industry partners. The role includes collaborative projects and preparation of publications, reports and presentations.

Your preparation checklist

Organize research evidence
Prepare a list of first-author papers and a brief account of your PhD research. The posting specifies a relevant PhD and at least three first-author papers; be ready to explain each paper’s contribution.
Build a manipulation project walkthrough
Select a relevant example, if available, and prepare a diagram showing the robotic task, perception and control methods, your contribution, results, and limitations. The role develops algorithms for robotic arms and embodied AI systems.
Practice a multimodal integration design
As a preparation exercise, sketch how vision, language, and force-sensor inputs could inform a robot action. Note how you would handle conflicting or unreliable inputs; the posting calls for helping integrate these modalities.
Prepare an experiment comparison
Outline how you would test robotic behavior in simulation and on hardware, including metrics, failure cases, and what would justify moving to a real-world trial. Both environments are named in the role.
Show implementation and hardware evidence
Gather concise examples of Python and C++ robotics work, ROS or MoveIt use, and hardware integration. Where applicable, explain your work with manipulators or mobile robots and how you operated or maintained equipment.
Rehearse research collaboration examples
Prepare one account of turning experimental results into a paper, report, or presentation, and one of working with research or industry collaborators. Distinguish your own contribution from the team’s.
Write questions for the research team
Ask which manipulation problems and platforms are priorities, how simulation and real-world experiments are balanced, and how the fellow would contribute to publications and collaborations. These clarify responsibilities stated in the posting, not promised projects.

Availability last checked: 4 Nov 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in Singapore. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…